v1.1.0 · 100% on-device · MIT

auge

Apple Vision from your terminal.

Point auge at any image. Get OCR text, classification labels, barcode payloads, and face bounding boxes back. One command, every analysis. The Vision framework is already on your Mac - auge gives you a UNIX CLI to it.

100% on-device No API keys Zero dependencies Pipe-friendly
$brew install Arthur-Ficial/tap/auge

Source: github.com/Arthur-Ficial/auge

Requires: macOS 10.15+ · Apple Silicon or Intel · No Xcode needed

Everything runs on your device. No network calls. No data leaves your Mac. Enforced at runtime by a URLProtocol guard that hard-exits any HTTP attempt.
Watch it work

One image in. Everything out.

The default mode is --all: OCR + classify + barcodes + faces, in one pass, on-device. Other tools make you pick. auge gives you everything.

Terminal — auge
$ auge --all photo.jpg === OCR === (no text) === CLASSIFY === animal: 92% cat: 92% feline: 92% mammal: 92% === BARCODES === (none) === FACES === 0 faces detected $ auge --all -o json scan.pdf # PDF natively, full structured output $ auge --all --md doc.png | apfel "summarize" $ auge --all --clipboard # NSPasteboard image input
Why auge

Zero APIs. Zero downloads. Zero cost.

Apple's Vision framework is already on your Mac, used by Photos and the Camera app. auge is a thin Swift wrapper that exposes it to the shell.

$0
Per request
No API keys, no usage tier, no per-token billing. Run it a million times.
100%
On-device
Zero network calls. Hard-enforced at runtime - any HTTP attempt aborts.
0
Dependencies
Pure Swift, no third-party packages. Vision ships with macOS.
187
Tests passing
Pure-Swift test runner, no XCTest. Same pattern as apfel.
Four analyses, one tool

What auge does.

Each analysis maps to a Vision framework request. Use them individually, or use --all for everything at once.

OCR

auge --ocr image.png

Text recognition via VNRecognizeTextRequest. Handles screenshots, scans, photos of signs, even handwriting. PDF input goes through PDFKit (text layer when present, rasterize for OCR otherwise).

$ auge --ocr --langs en-US,de-DE doc.png
Hello World
Grüß Gott

Classify

auge --classify image.jpg

Image classification via VNClassifyImageRequest. 1000+ categories with confidence percentages. Animals, vehicles, objects, scenes, materials - the same model Photos uses.

$ auge --classify cat.jpg
animal: 92%
cat: 92%
feline: 92%

Barcodes

auge --barcode image.png

Barcode and QR detection via VNDetectBarcodesRequest. Returns the decoded payload plus the symbology (QR, EAN, Code128, PDF417, Aztec, and more).

$ auge --barcode qr.png
[QR] http://en.m.wikipedia.org

Faces

auge --faces image.jpg

Face detection via VNDetectFaceRectanglesRequest. Counts faces and returns normalized bounding boxes (x, y, width, height) you can draw over the image.

$ auge --faces -o json group.jpg
{"count":7,"faces":[
  {"x":0.21,"y":0.38,"w":0.49,"h":0.38},
  ...
]}
How it works

Apple's vision engine. UNIX semantics.

auge is a thin Swift CLI over the Vision framework. The model lives in macOS - auge just gives you a way to call it from the shell.

1

Image in (any source)

File path, stdin pipe (cat x.png | auge --ocr), --clipboard for NSPasteboard, or PDF (PDFKit handles text layer or rasterizes pages).

2

Vision request, on-device

auge calls the matching VN…Request on your Mac's Apple Silicon (or Intel) GPU/Neural Engine. The model is shipped with macOS - no download, no API.

3

Structured output

Plain text, Markdown, JSON, NDJSON, or compact short-key JSON. Pipe it to jq, apfel, llm, or any UNIX tool.

The data flow

image / pdf / clipboard / stdin
NetworkGuard (URLProtocol)
Vision framework (on-device)
JSON / Plain / Markdown / NDJSON → stdout

NetworkGuard registers a URLProtocol that intercepts every http / https / ws / wss request inside the auge process and exits non-zero with a stderr message. If a future dependency, plugin, or accidental bug ever tries to reach the network, auge dies before the call lands. Belt and suspenders.

Capabilities

Everything auge supports.

One binary, every mode, every common image format, dozens of languages, every major barcode standard. All on-device.

Input formats

  • PNG · JPEG · HEIC · HEIF
  • TIFF · BMP · GIF
  • PDF (multi-page, text-layer or rasterize)
  • NSPasteboard image (--clipboard)
  • Stdin pipe (file paths or bytes)

Analyses

  • --ocr Text recognition (VNRecognizeTextRequest)
  • --classify 1000+ labels (VNClassifyImageRequest)
  • --barcode 12 symbologies (VNDetectBarcodesRequest)
  • --faces Bounding boxes (VNDetectFaceRectanglesRequest)
  • --all Every analysis in one pass

Output formats

  • plain Default; structured per mode
  • json Pretty-printed structured output
  • --compact Single-line JSON
  • ndjson One JSON record per line
  • md GitHub-flavored Markdown

Barcode symbologies

  • QR · Aztec · DataMatrix
  • EAN-13 · EAN-8 · UPC-A · UPC-E
  • Code 128 · Code 39 · Code 93
  • PDF417 · ITF (Interleaved 2 of 5)
  • Plus URL/vCard/WiFi schema decoding

OCR languages (BCP-47)

  • Latin script: en, de, fr, es, it, pt, nl, sv, da, nb, fi, pl, cs, hu, tr, vi, cy, la
  • CJK: zh-Hans, zh-Hant, ja, ko
  • RTL: ar, he, fa, ur
  • Other: ru, uk, sr-Cyrl, sr-Latn, el, hi, th, id
  • Mix any of them with --langs a,b,c

OCR controls

  • --langs BCP-47 hints (priority order)
  • --enhance Upscale tiny inputs before OCR
  • --dpi PDF rasterization 72-600 (default 200)
  • --prefer-embedded Use PDF text layer if present
  • --clean FoundationModels post-pass (macOS 26+)

Privacy guarantees

  • 100% on-device · No API keys · No telemetry
  • NetworkGuard URLProtocol blocks every http/https/ws/wss at runtime
  • Zero third-party Swift dependencies
  • No model downloads (Vision ships with macOS)
  • MIT license

Platform & build

  • macOS 10.15+ (full set on macOS 26 Tahoe)
  • Apple Silicon or Intel
  • Builds with Command Line Tools - no Xcode
  • Swift 6.3 strict concurrency
  • Hand-rolled CLI - no swift-argument-parser
Real documents, real auge

Public-domain corpus, real outcomes.

Every example below is processed by running the real auge binary (v1.1.0) on a public-domain document from Wikimedia Commons - at build time, on a Mac, with no network. Each card shows what auge produced: structured OCR text, classification labels with confidence, barcode payloads, face bounding boxes overlaid on the image, plus the full JSON output, on-device. Nothing is mocked or post-processed.

61examples below
14with face detection
26with OCR text extraction
10with QR / barcode decoding
61with classification labels
16distinct languages OCR'd
The transistor inventors - Bell Labs, 1948
face 1
face 2
face 3
faces 1948 Public Domain (Bell Labs PR photo, PD-US-no-notice)

The transistor inventors - Bell Labs, 1948

John Bardeen, William Shockley, Walter Brattain - the three Bell Labs scientists who invented the transistor in December 1947, photographed the next year. Every CPU on Earth descends from this workbench. auge detects all three faces.

$ auge --all bell-labs-transistor.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
people
81%
adult
81%
clothing
71%
necktie
67%
cord
49%
suit
49%
structure
35%
furniture
35%
table
34%
tableware
26%
Barcodes none
No barcodes or QR codes detected.
Faces 3 detected
3 faces located. Bounding boxes overlaid on the image above.
face 1: x=0.273 y=0.700 w=0.123 h=0.155
face 2: x=0.442 y=0.495 w=0.119 h=0.149
face 3: x=0.609 y=0.651 w=0.129 h=0.162
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/bell-labs-transistor.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8076185584068298,
          "label": "people"
        },
        {
          "confidence": 0.8076171875,
          "label": "adult"
        },
        {
          "confidence": 0.7065683007240295,
          "label": "clothing"
        },
        {
          "confidence": 0.67236328125,
          "label": "necktie"
        },
        {
          "confidence": 0.4892578125,
          "label": "cord"
        },
        {
          "confidence": 0.4873046875,
          "label": "suit"
        },
        {
          "confidence": 0.3476963937282562,
          "label": "structure"
        },
        {
          "confidence": 0.3450329601764679,
          "label": "furniture"
        },
        {
          "confidence": 0.344970703125,
          "label": "table"
        },
        {
          "confidence": 0.26153215765953064,
          "label": "tableware"
        }
      ]
    },
    "faces": {
      "count": 3,
      "faces": [
        {
          "height": 0.1548747420310974,
          "width": 0.12341581284999847,
          "x": 0.27318617701530457,
          "y": 0.7003240585327148
        },
        {
          "height": 0.14936710894107819,
          "width": 0.11902691423892975,
          "x": 0.44163864850997925,
          "y": 0.4954269528388977
        },
        {
          "height": 0.16231800615787506,
          "width": 0.12934716045856476,
          "x": 0.6089426279067993,
          "y": 0.6511827111244202
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Earthrise - Apollo 8, 24 December 1968
classify 1968 Public Domain (NASA work)

Earthrise - Apollo 8, 24 December 1968

Bill Anders's photograph of Earth rising over the Moon, taken from Apollo 8 on Christmas Eve 1968. The most-reproduced photo of Earth ever taken. Galen Rowell called it 'the most influential environmental photograph ever taken.'

$ auge --all earthrise.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
outdoor
58%
celestial_body
57%
moon
57%
night_sky
34%
sky
34%
structure
20%
rocks
20%
liquid
16%
water
16%
frozen
16%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/earthrise.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.5810793042182922,
          "label": "outdoor"
        },
        {
          "confidence": 0.5659338235855103,
          "label": "celestial_body"
        },
        {
          "confidence": 0.56591796875,
          "label": "moon"
        },
        {
          "confidence": 0.3359375,
          "label": "night_sky"
        },
        {
          "confidence": 0.3359375,
          "label": "sky"
        },
        {
          "confidence": 0.1994629055261612,
          "label": "structure"
        },
        {
          "confidence": 0.199462890625,
          "label": "rocks"
        },
        {
          "confidence": 0.16483356058597565,
          "label": "liquid"
        },
        {
          "confidence": 0.16483356058597565,
          "label": "water"
        },
        {
          "confidence": 0.1648075133562088,
          "label": "frozen"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Banquet of the Officers of the St Hadrian Civic Guard - Frans Hals
face 1
face 2
face 3
face 4
face 5
face 6
face 7
face 8
face 9
face 10
face 11
faces 1627 Public Domain (Hals died 1666)

Banquet of the Officers of the St Hadrian Civic Guard - Frans Hals

Twelve officers at a banquet, Haarlem 1627. Frans Hals founded the Dutch group portrait genre - eyes alive, faces individual. A perfect multi-face benchmark.

$ auge --all hals-banquet.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
84%
painting
84%
people
83%
adult
82%
material
55%
textile
55%
crowd
40%
structure
33%
conveyance
33%
portal
33%
Barcodes none
No barcodes or QR codes detected.
Faces 11 detected
11 faces located. Bounding boxes overlaid on the image above.
face 1: x=0.102 y=0.641 w=0.054 h=0.078
face 2: x=0.253 y=0.614 w=0.046 h=0.067
face 3: x=0.426 y=0.429 w=0.049 h=0.072
face 4: x=0.861 y=0.440 w=0.050 h=0.072
face 5: x=0.592 y=0.624 w=0.039 h=0.057
face 6: x=0.575 y=0.421 w=0.047 h=0.068
face 7: x=0.739 y=0.422 w=0.051 h=0.074
face 8: x=0.401 y=0.555 w=0.040 h=0.058
face 9: x=0.858 y=0.636 w=0.049 h=0.071
face 10: x=0.254 y=0.432 w=0.044 h=0.064
face 11: x=0.689 y=0.694 w=0.044 h=0.064
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/hals-banquet.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8378981947898865,
          "label": "art"
        },
        {
          "confidence": 0.837890625,
          "label": "painting"
        },
        {
          "confidence": 0.82733553647995,
          "label": "people"
        },
        {
          "confidence": 0.82080078125,
          "label": "adult"
        },
        {
          "confidence": 0.5494153499603271,
          "label": "material"
        },
        {
          "confidence": 0.54931640625,
          "label": "textile"
        },
        {
          "confidence": 0.404052734375,
          "label": "crowd"
        },
        {
          "confidence": 0.3328096568584442,
          "label": "structure"
        },
        {
          "confidence": 0.3322772681713104,
          "label": "conveyance"
        },
        {
          "confidence": 0.3322772681713104,
          "label": "portal"
        }
      ]
    },
    "faces": {
      "count": 11,
      "faces": [
        {
          "height": 0.07813137024641037,
          "width": 0.05371531844139099,
          "x": 0.10235835611820221,
          "y": 0.6411809921264648
        },
        {
          "height": 0.06691820174455643,
          "width": 0.04600626602768898,
          "x": 0.25293394923210144,
          "y": 0.6139742732048035
        },
        {
          "height": 0.07153363525867462,
          "width": 0.04917937144637108,
          "x": 0.42617541551589966,
          "y": 0.42902836203575134
        },
        {
          "height": 0.07247375696897507,
          "width": 0.04982570931315422,
          "x": 0.860861599445343,
          "y": 0.4400157630443573
        },
        {
          "height": 0.05714447796344757,
          "width": 0.03928682953119278,
          "x": 0.592441201210022,
          "y": 0.6244773864746094
        },
        {
          "height": 0.06817885488271713,
          "width": 0.04687296599149704,
          "x": 0.5745560526847839,
          "y": 0.4207250773906708
        },
        {
          "height": 0.07369133085012436,
          "width": 0.05066279321908951,
          "x": 0.7387236952781677,
          "y": 0.4220446050167084
        },
        {
          "height": 0.05780195817351341,
          "width": 0.03973884508013725,
          "x": 0.40064874291419983,
          "y": 0.555475115776062
        },
        {
          "height": 0.07082018256187439,
          "width": 0.04868887737393379,
          "x": 0.8580805659294128,
          "y": 0.6364502906799316
        },
        {
          "height": 0.06449496746063232,
          "width": 0.04434029012918472,
          "x": 0.25403550267219543,
          "y": 0.4322361648082733
        },
        {
          "height": 0.06371127814054489,
          "width": 0.043801501393318176,
          "x": 0.6891831755638123,
          "y": 0.6943004727363586
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Multilingual sign - Macau
multilingual 2010 Public Domain

Multilingual sign - Macau

Chinese, Portuguese, and English on a single Macau sign. With --langs, auge handles all three scripts in one OCR pass.

$ auge --all --langs en-US,pt-PT,zh-Hant --enhance multilingual-sign.png
A OCR 6 lines · 77 chars
Hotéis
Hotels
Informações
Information
Agências de Viagens
Travel Agencies
# Classify 10 labels
material
37%
raw_glass
37%
structure
28%
sign
28%
fence
13%
art
7%
decoration
7%
frame
7%
conveyance
5%
easel
5%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/multilingual-sign.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.37218672037124634,
          "label": "material"
        },
        {
          "confidence": 0.3720703125,
          "label": "raw_glass"
        },
        {
          "confidence": 0.278434693813324,
          "label": "structure"
        },
        {
          "confidence": 0.27539098262786865,
          "label": "sign"
        },
        {
          "confidence": 0.125732421875,
          "label": "fence"
        },
        {
          "confidence": 0.07118187844753265,
          "label": "art"
        },
        {
          "confidence": 0.06570933014154434,
          "label": "decoration"
        },
        {
          "confidence": 0.065673828125,
          "label": "frame"
        },
        {
          "confidence": 0.05234569311141968,
          "label": "conveyance"
        },
        {
          "confidence": 0.05126953125,
          "label": "easel"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "Hotéis",
        "Hotels",
        "Informações",
        "Information",
        "Agências de Viagens",
        "Travel Agencies"
      ],
      "text": "Hotéis\nHotels\nInformações\nInformation\nAgências de Viagens\nTravel Agencies"
    }
  }
}
Wikimedia source
QR code - what's behind this link?
barcode 2026 CC0 (locally generated with qrencode)

QR code - what's behind this link?

Most QR codes you scan in the wild, you scan blind. auge decodes the URL on-device first - so you can read the destination before your phone does. Some links are surprises. Some are jokes. This one is a 38-year-old joke that still works.

$ auge --all qr-mystery.png
A OCR no text
No text detected in this image.
# Classify 10 labels
art
7%
illustrations
7%
document
2%
chart
2%
diagram
2%
printed_page
2%
structure
1%
sign
1%
decoration
1%
frame
1%
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-mystery.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
          "symbology": "QR"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.06788385659456253,
          "label": "art"
        },
        {
          "confidence": 0.06787109375,
          "label": "illustrations"
        },
        {
          "confidence": 0.023224662989377975,
          "label": "document"
        },
        {
          "confidence": 0.021240234375,
          "label": "chart"
        },
        {
          "confidence": 0.021240234375,
          "label": "diagram"
        },
        {
          "confidence": 0.01708984375,
          "label": "printed_page"
        },
        {
          "confidence": 0.014018773101270199,
          "label": "structure"
        },
        {
          "confidence": 0.013677000068128109,
          "label": "sign"
        },
        {
          "confidence": 0.011230524629354477,
          "label": "decoration"
        },
        {
          "confidence": 0.01123046875,
          "label": "frame"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Anatomy Lesson of Dr Nicolaes Tulp - Rembrandt, 1632
face 1
face 2
face 3
face 4
face 5
face 6
face 7
face 8
faces 1632 Public Domain

The Anatomy Lesson of Dr Nicolaes Tulp - Rembrandt, 1632

Rembrandt at 26, painting the Amsterdam Surgeons' Guild. Eight faces around a dissection - intimate, geometric, masterful. auge detects each one.

$ auge --all rembrandt-anatomy.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
people
61%
adult
61%
art
39%
painting
39%
clothing
27%
hat
27%
headgear
27%
fedora
27%
structure
20%
wood_processed
18%
Barcodes none
No barcodes or QR codes detected.
Faces 8 detected
8 faces located. Bounding boxes overlaid on the image above.
face 1: x=0.304 y=0.610 w=0.061 h=0.081
face 2: x=0.350 y=0.453 w=0.056 h=0.074
face 3: x=0.160 y=0.477 w=0.059 h=0.078
face 4: x=0.409 y=0.490 w=0.065 h=0.086
face 5: x=0.253 y=0.775 w=0.060 h=0.079
face 6: x=0.593 y=0.599 w=0.065 h=0.087
face 7: x=0.476 y=0.672 w=0.059 h=0.079
face 8: x=0.050 y=0.491 w=0.051 h=0.068
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/rembrandt-anatomy.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6130247712135315,
          "label": "people"
        },
        {
          "confidence": 0.6123046875,
          "label": "adult"
        },
        {
          "confidence": 0.3891679346561432,
          "label": "art"
        },
        {
          "confidence": 0.38916015625,
          "label": "painting"
        },
        {
          "confidence": 0.2665797173976898,
          "label": "clothing"
        },
        {
          "confidence": 0.2663705050945282,
          "label": "hat"
        },
        {
          "confidence": 0.2663705050945282,
          "label": "headgear"
        },
        {
          "confidence": 0.26513671875,
          "label": "fedora"
        },
        {
          "confidence": 0.20316724479198456,
          "label": "structure"
        },
        {
          "confidence": 0.178955078125,
          "label": "wood_processed"
        }
      ]
    },
    "faces": {
      "count": 8,
      "faces": [
        {
          "height": 0.08094076067209244,
          "width": 0.060958512127399445,
          "x": 0.304284930229187,
          "y": 0.6098484992980957
        },
        {
          "height": 0.07414423674345016,
          "width": 0.05583987757563591,
          "x": 0.3498956561088562,
          "y": 0.4533364474773407
        },
        {
          "height": 0.078190878033638,
          "width": 0.058887504041194916,
          "x": 0.15966328978538513,
          "y": 0.47676825523376465
        },
        {
          "height": 0.0859607383608818,
          "width": 0.0647391825914383,
          "x": 0.40939679741859436,
          "y": 0.48952704668045044
        },
        {
          "height": 0.07902317494153976,
          "width": 0.059514325112104416,
          "x": 0.2526022791862488,
          "y": 0.7752046585083008
        },
        {
          "height": 0.08653508871793747,
          "width": 0.0651717334985733,
          "x": 0.5929089784622192,
          "y": 0.5990678668022156
        },
        {
          "height": 0.07869797199964523,
          "width": 0.059269409626722336,
          "x": 0.475503534078598,
          "y": 0.6718984246253967
        },
        {
          "height": 0.06783340871334076,
          "width": 0.05108703300356865,
          "x": 0.05027836561203003,
          "y": 0.49100106954574585
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Blue Marble - Apollo 17, 1972
classify 1972 Public Domain (NASA work)

The Blue Marble - Apollo 17, 1972

Apollo 17, 7 December 1972. The crew, the last humans to leave Low Earth Orbit, took the only fully-illuminated photo of Earth made by an astronaut. Vision classifies 'astronomy / planet / globe'.

$ auge --all blue-marble.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
outdoor
76%
celestial_body
76%
moon
76%
liquid
11%
water
11%
frozen
11%
ice
11%
sky
5%
night_sky
5%
snow
2%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/blue-marble.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7563506364822388,
          "label": "outdoor"
        },
        {
          "confidence": 0.7563477158546448,
          "label": "celestial_body"
        },
        {
          "confidence": 0.75634765625,
          "label": "moon"
        },
        {
          "confidence": 0.10739319026470184,
          "label": "liquid"
        },
        {
          "confidence": 0.10739319026470184,
          "label": "water"
        },
        {
          "confidence": 0.10721195489168167,
          "label": "frozen"
        },
        {
          "confidence": 0.107177734375,
          "label": "ice"
        },
        {
          "confidence": 0.0524904727935791,
          "label": "sky"
        },
        {
          "confidence": 0.052490234375,
          "label": "night_sky"
        },
        {
          "confidence": 0.020263671875,
          "label": "snow"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Vatican ATM with Latin instructions
multilingual 2014 Public Domain

Vatican ATM with Latin instructions

The only ATM in the world with a Latin user interface. auge OCRs medieval-screen Latin: 'Inserito scidulam quaeso ut faciundam cognoscas rationem'.

$ auge --all --langs la,it-IT vatican-atm.jpg
A OCR 6 lines · 56 chars
INSERITO
SCIDULAM
QUAESO UT
FACTUNDAM
COGNOSCAS
RATIONEM
# Classify 10 labels
machine
61%
atm
61%
keypad
12%
consumer_electronics
2%
television
2%
structure
2%
conveyance
2%
elevator
2%
computer
1%
computer_monitor
1%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/vatican-atm.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6081007122993469,
          "label": "machine"
        },
        {
          "confidence": 0.60791015625,
          "label": "atm"
        },
        {
          "confidence": 0.11865234375,
          "label": "keypad"
        },
        {
          "confidence": 0.021320780739188194,
          "label": "consumer_electronics"
        },
        {
          "confidence": 0.02099609375,
          "label": "television"
        },
        {
          "confidence": 0.017412301152944565,
          "label": "structure"
        },
        {
          "confidence": 0.017273560166358948,
          "label": "conveyance"
        },
        {
          "confidence": 0.01708984375,
          "label": "elevator"
        },
        {
          "confidence": 0.010501927696168423,
          "label": "computer"
        },
        {
          "confidence": 0.010498046875,
          "label": "computer_monitor"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "INSERITO",
        "SCIDULAM",
        "QUAESO UT",
        "FACTUNDAM",
        "COGNOSCAS",
        "RATIONEM"
      ],
      "text": "INSERITO\nSCIDULAM\nQUAESO UT\nFACTUNDAM\nCOGNOSCAS\nRATIONEM"
    }
  }
}
Wikimedia source
Abraham Lincoln, 1863
face 1
faces 1863 Public Domain

Abraham Lincoln, 1863

A 19th-century portrait. auge detects the face (with bounding box), classifies the image, and OCRs any visible text.

$ auge --all lincoln.jpg
A OCR no text
No text detected in this image.
# Classify 8 labels
clothing
76%
suit
73%
people
69%
adult
69%
bowtie
52%
necktie
26%
tuxedo
18%
military_uniform
3%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.221 y=0.382 w=0.486 h=0.377
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/lincoln.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.764702558517456,
          "label": "clothing"
        },
        {
          "confidence": 0.7314453125,
          "label": "suit"
        },
        {
          "confidence": 0.6875183582305908,
          "label": "people"
        },
        {
          "confidence": 0.6875,
          "label": "adult"
        },
        {
          "confidence": 0.51953125,
          "label": "bowtie"
        },
        {
          "confidence": 0.258544921875,
          "label": "necktie"
        },
        {
          "confidence": 0.184326171875,
          "label": "tuxedo"
        },
        {
          "confidence": 0.033203125,
          "label": "military_uniform"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.3769800066947937,
          "width": 0.4856562614440918,
          "x": 0.22074520587921143,
          "y": 0.38192880153656006
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Apollo 11 crew portrait, 1969
face 1
face 2
face 3
faces 1969 Public Domain (NASA work)

Apollo 11 crew portrait, 1969

Armstrong, Collins, Aldrin in pre-flight portrait. NASA published this and every other crew photo into the public domain - try OCR-ing the name tags on the spacesuits. auge does both modes in one pass.

$ auge --all apollo-11-crew.jpg
A OCR 6 lines · 38 chars
NACA
ARMSTRONG
NA?
COLLINS
NASA
ALDRIN
# Classify 10 labels
people
75%
adult
75%
structure
22%
arch
22%
recreation
10%
sport
10%
fencing_sport
9%
clothing
8%
military_uniform
6%
jacket
6%
Barcodes none
No barcodes or QR codes detected.
Faces 3 detected
3 faces located. Bounding boxes overlaid on the image above.
face 1: x=0.673 y=0.496 w=0.092 h=0.118
face 2: x=0.441 y=0.698 w=0.089 h=0.113
face 3: x=0.210 y=0.511 w=0.095 h=0.121
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/apollo-11-crew.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7475586533546448,
          "label": "people"
        },
        {
          "confidence": 0.74755859375,
          "label": "adult"
        },
        {
          "confidence": 0.21655291318893433,
          "label": "structure"
        },
        {
          "confidence": 0.216552734375,
          "label": "arch"
        },
        {
          "confidence": 0.09518646448850632,
          "label": "recreation"
        },
        {
          "confidence": 0.09518645703792572,
          "label": "sport"
        },
        {
          "confidence": 0.0947265625,
          "label": "fencing_sport"
        },
        {
          "confidence": 0.07742087543010712,
          "label": "clothing"
        },
        {
          "confidence": 0.063720703125,
          "label": "military_uniform"
        },
        {
          "confidence": 0.0625,
          "label": "jacket"
        }
      ]
    },
    "faces": {
      "count": 3,
      "faces": [
        {
          "height": 0.11763617396354675,
          "width": 0.09245467931032181,
          "x": 0.6728711128234863,
          "y": 0.4963274598121643
        },
        {
          "height": 0.11271987855434418,
          "width": 0.08859077841043472,
          "x": 0.44056910276412964,
          "y": 0.6976574063301086
        },
        {
          "height": 0.12080634385347366,
          "width": 0.0949462354183197,
          "x": 0.21001489460468292,
          "y": 0.5114022493362427
        }
      ]
    },
    "ocr": {
      "lines": [
        "NACA",
        "ARMSTRONG",
        "NA?",
        "COLLINS",
        "NASA",
        "ALDRIN"
      ],
      "text": "NACA\nARMSTRONG\nNA?\nCOLLINS\nNASA\nALDRIN"
    }
  }
}
Wikimedia source
QR code - Wikipedia URL
barcode 2024 CC0 (algorithmic)

QR code - Wikipedia URL

A QR code that decodes to Wikipedia's mobile URL. auge --all decodes the barcode and runs every other analysis on the same image, on-device, no camera, no network.

$ auge --all qr-wikipedia.png
A OCR no text
No text detected in this image.
# Classify 10 labels
art
9%
illustrations
9%
document
4%
printed_page
3%
chart
3%
diagram
3%
structure
2%
sign
2%
decoration
1%
frame
1%
Barcodes 1 found
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-wikipedia.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "http://en.m.wikipedia.org",
          "symbology": "QR"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.08667552471160889,
          "label": "art"
        },
        {
          "confidence": 0.086669921875,
          "label": "illustrations"
        },
        {
          "confidence": 0.035830721259117126,
          "label": "document"
        },
        {
          "confidence": 0.033203125,
          "label": "printed_page"
        },
        {
          "confidence": 0.025634765625,
          "label": "chart"
        },
        {
          "confidence": 0.025634765625,
          "label": "diagram"
        },
        {
          "confidence": 0.018115323036909103,
          "label": "structure"
        },
        {
          "confidence": 0.018074041232466698,
          "label": "sign"
        },
        {
          "confidence": 0.010986343957483768,
          "label": "decoration"
        },
        {
          "confidence": 0.010986328125,
          "label": "frame"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Pillars of Creation - Hubble, 2014
classify 2014 Public Domain (NASA / ESA)

Pillars of Creation - Hubble, 2014

Three columns of cold gas and dust in the Eagle Nebula, 6,500 light-years from Earth. Hubble first imaged them in 1995; this is the 2014 reshoot at higher resolution. Vision classifies 'astronomy / nebula / sky'.

$ auge --all pillars-of-creation.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
animal
21%
fish
21%
seahorse
21%
liquid
13%
water
13%
underwater
13%
outdoor
9%
sky
9%
night_sky
9%
fire
9%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/pillars-of-creation.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.2114260196685791,
          "label": "animal"
        },
        {
          "confidence": 0.21142578125,
          "label": "fish"
        },
        {
          "confidence": 0.21142578125,
          "label": "seahorse"
        },
        {
          "confidence": 0.1267145425081253,
          "label": "liquid"
        },
        {
          "confidence": 0.1267145425081253,
          "label": "water"
        },
        {
          "confidence": 0.126708984375,
          "label": "underwater"
        },
        {
          "confidence": 0.09388422966003418,
          "label": "outdoor"
        },
        {
          "confidence": 0.09387935698032379,
          "label": "sky"
        },
        {
          "confidence": 0.091552734375,
          "label": "night_sky"
        },
        {
          "confidence": 0.0855553075671196,
          "label": "fire"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Mona Lisa - Leonardo da Vinci, 1503
face 1
historical 1503 Public Domain (Leonardo died 1519)

Mona Lisa - Leonardo da Vinci, 1503

Probably the most-recognized painting on Earth. auge detects her face and classifies the image - a useful sanity check for what Vision considers a 'painting' vs 'portrait'.

$ auge --all mona-lisa.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
85%
painting
85%
people
44%
adult
44%
clothing
6%
cloak
6%
outdoor
3%
sky
3%
cloudy
3%
illustrations
2%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.309 y=0.638 w=0.285 h=0.188
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/mona-lisa.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8476563096046448,
          "label": "art"
        },
        {
          "confidence": 0.84765625,
          "label": "painting"
        },
        {
          "confidence": 0.44165050983428955,
          "label": "people"
        },
        {
          "confidence": 0.441650390625,
          "label": "adult"
        },
        {
          "confidence": 0.0583508238196373,
          "label": "clothing"
        },
        {
          "confidence": 0.058349609375,
          "label": "cloak"
        },
        {
          "confidence": 0.02913166582584381,
          "label": "outdoor"
        },
        {
          "confidence": 0.0274445079267025,
          "label": "sky"
        },
        {
          "confidence": 0.025390625,
          "label": "cloudy"
        },
        {
          "confidence": 0.0244140625,
          "label": "illustrations"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.18824096024036407,
          "width": 0.2847144603729248,
          "x": 0.3091900646686554,
          "y": 0.6384798288345337
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Domestic cat
classify 2010 CC0

Domestic cat

Point auge at any image. It returns OCR text (none here), the top classification labels with confidence, faces, and barcodes - all in one pass.

$ auge --all --top 8 cat.jpg
A OCR no text
No text detected in this image.
# Classify 8 labels
animal
92%
cat
92%
feline
92%
mammal
92%
adult_cat
92%
frozen
28%
liquid
28%
snow
28%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/cat.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.9150391817092896,
          "label": "animal"
        },
        {
          "confidence": 0.9150391817092896,
          "label": "cat"
        },
        {
          "confidence": 0.9150391817092896,
          "label": "feline"
        },
        {
          "confidence": 0.9150391817092896,
          "label": "mammal"
        },
        {
          "confidence": 0.9150390625,
          "label": "adult_cat"
        },
        {
          "confidence": 0.284423828125,
          "label": "frozen"
        },
        {
          "confidence": 0.284423828125,
          "label": "liquid"
        },
        {
          "confidence": 0.284423828125,
          "label": "snow"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Great Wave off Kanagawa - Hokusai, 1831
classify 1831 Public Domain

The Great Wave off Kanagawa - Hokusai, 1831

Hokusai's signature woodblock print from his Thirty-six Views of Mount Fuji. The most reproduced Japanese artwork ever made. Classify pulls 'art / painting / wave / sea'.

$ auge --all hokusai-wave.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
69%
illustrations
69%
painting
18%
document
5%
printed_page
5%
material
4%
textile
4%
liquid
3%
water
3%
underwater
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/hokusai-wave.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6895982623100281,
          "label": "art"
        },
        {
          "confidence": 0.68896484375,
          "label": "illustrations"
        },
        {
          "confidence": 0.180908203125,
          "label": "painting"
        },
        {
          "confidence": 0.0546172596514225,
          "label": "document"
        },
        {
          "confidence": 0.05419921875,
          "label": "printed_page"
        },
        {
          "confidence": 0.03515634685754776,
          "label": "material"
        },
        {
          "confidence": 0.03515625,
          "label": "textile"
        },
        {
          "confidence": 0.029265359044075012,
          "label": "liquid"
        },
        {
          "confidence": 0.029265359044075012,
          "label": "water"
        },
        {
          "confidence": 0.029052734375,
          "label": "underwater"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Girl with a Pearl Earring - Vermeer, 1665
face 1
faces 1665 Public Domain

Girl with a Pearl Earring - Vermeer, 1665

The 'Mona Lisa of the North'. The bounding box should fit her tilted face cleanly - a good test of normalized-coordinate accuracy.

$ auge --all vermeer-pearl.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
65%
painting
65%
people
28%
adult
28%
clothing
8%
headgear
8%
child
2%
illustrations
2%
cloak
2%
hat
1%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.244 y=0.415 w=0.316 h=0.267
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/vermeer-pearl.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6513673067092896,
          "label": "art"
        },
        {
          "confidence": 0.6513671875,
          "label": "painting"
        },
        {
          "confidence": 0.28276556730270386,
          "label": "people"
        },
        {
          "confidence": 0.28271484375,
          "label": "adult"
        },
        {
          "confidence": 0.07832333445549011,
          "label": "clothing"
        },
        {
          "confidence": 0.07825804501771927,
          "label": "headgear"
        },
        {
          "confidence": 0.02001953125,
          "label": "child"
        },
        {
          "confidence": 0.01904296875,
          "label": "illustrations"
        },
        {
          "confidence": 0.018798828125,
          "label": "cloak"
        },
        {
          "confidence": 0.011601286008954048,
          "label": "hat"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.26685962080955505,
          "width": 0.3160618841648102,
          "x": 0.24406862258911133,
          "y": 0.41480395197868347
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
New-York Daily Tribune, 1865
historical 1865 Public Domain

New-York Daily Tribune, 1865

A real 1865 newspaper PDF from the Internet Archive. auge handles PDF natively via PDFKit and runs every other analysis on each page.

$ auge --all ny-tribune-1865.pdf
A OCR 4 lines · 56 chars
EUROPE.
Dem-Mork
THE SEVEN-TAIOTIES.
THE DISASTER AT BRA
# Classify 8 labels
document
66%
printed_page
66%
newspaper
15%
book
11%
screenshot
2%
handwriting
2%
art
2%
illustrations
2%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/ny-tribune-1865.pdf",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6634981632232666,
          "label": "document"
        },
        {
          "confidence": 0.6630859375,
          "label": "printed_page"
        },
        {
          "confidence": 0.14599609375,
          "label": "newspaper"
        },
        {
          "confidence": 0.109130859375,
          "label": "book"
        },
        {
          "confidence": 0.022216796875,
          "label": "screenshot"
        },
        {
          "confidence": 0.020751953125,
          "label": "handwriting"
        },
        {
          "confidence": 0.015136785805225372,
          "label": "art"
        },
        {
          "confidence": 0.01513671875,
          "label": "illustrations"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "EUROPE.",
        "Dem-Mork",
        "THE SEVEN-TAIOTIES.",
        "THE DISASTER AT BRA"
      ],
      "text": "EUROPE.\nDem-Mork\nTHE SEVEN-TAIOTIES.\nTHE DISASTER AT BRA"
    }
  }
}
Wikimedia source
Welsh-English no-parking sign, Wales
multilingual 2010 Public Domain

Welsh-English no-parking sign, Wales

Welsh signage law requires every public sign to be bilingual. auge --langs cy,en-GB OCRs both 'Dim Parcio' and 'No Parking' in one pass.

$ auge --all --langs cy,en-GB welsh-english-sign.jpg
A OCR 7 lines · 67 chars
Dim Parcio
Mae Angen
Mynediad
24 Awr
No Parking
24h Access
Required
# Classify 10 labels
structure
86%
sign
86%
material
43%
brick
43%
street_sign
37%
conveyance
11%
portal
11%
window
11%
raw_glass
5%
stairs
2%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/welsh-english-sign.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8647416830062866,
          "label": "structure"
        },
        {
          "confidence": 0.8647142648696899,
          "label": "sign"
        },
        {
          "confidence": 0.4311799108982086,
          "label": "material"
        },
        {
          "confidence": 0.43115234375,
          "label": "brick"
        },
        {
          "confidence": 0.370361328125,
          "label": "street_sign"
        },
        {
          "confidence": 0.112587109208107,
          "label": "conveyance"
        },
        {
          "confidence": 0.11255455762147903,
          "label": "portal"
        },
        {
          "confidence": 0.112548828125,
          "label": "window"
        },
        {
          "confidence": 0.050537109375,
          "label": "raw_glass"
        },
        {
          "confidence": 0.020751953125,
          "label": "stairs"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "Dim Parcio",
        "Mae Angen",
        "Mynediad",
        "24 Awr",
        "No Parking",
        "24h Access",
        "Required"
      ],
      "text": "Dim Parcio\nMae Angen\nMynediad\n24 Awr\nNo Parking\n24h Access\nRequired"
    }
  }
}
Wikimedia source
Sombrero Galaxy (M104) - Hubble
classify 2003 Public Domain (NASA / ESA)

Sombrero Galaxy (M104) - Hubble

M104 in Virgo, 28 million light-years away. The dust lane edge-on gives it the namesake hat shape. Vision classifies 'galaxy / astronomy'.

$ auge --all sombrero-galaxy.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
outdoor
83%
sky
83%
night_sky
83%
aurora
5%
cloudy
3%
liquid
3%
water
3%
frozen
3%
snow
3%
water_body
2%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/sombrero-galaxy.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8325209021568298,
          "label": "outdoor"
        },
        {
          "confidence": 0.8325209021568298,
          "label": "sky"
        },
        {
          "confidence": 0.83251953125,
          "label": "night_sky"
        },
        {
          "confidence": 0.048095703125,
          "label": "aurora"
        },
        {
          "confidence": 0.02978515625,
          "label": "cloudy"
        },
        {
          "confidence": 0.026948798447847366,
          "label": "liquid"
        },
        {
          "confidence": 0.026948798447847366,
          "label": "water"
        },
        {
          "confidence": 0.02587890625,
          "label": "frozen"
        },
        {
          "confidence": 0.02587890625,
          "label": "snow"
        },
        {
          "confidence": 0.016758495941758156,
          "label": "water_body"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Marie Curie, c. 1920
face 1
faces 1920 Public Domain

Marie Curie, c. 1920

The only person ever to win Nobel Prizes in two different sciences. auge detects the face and classifies the studio portrait.

$ auge --all marie-curie.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
adult
89%
people
89%
clothing
19%
suit
18%
necktie
13%
art
9%
decoration
9%
jewelry
9%
military_uniform
3%
bowtie
2%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.227 y=0.425 w=0.423 h=0.311
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/marie-curie.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8916015625,
          "label": "adult"
        },
        {
          "confidence": 0.8916015625,
          "label": "people"
        },
        {
          "confidence": 0.1919054090976715,
          "label": "clothing"
        },
        {
          "confidence": 0.181396484375,
          "label": "suit"
        },
        {
          "confidence": 0.12841796875,
          "label": "necktie"
        },
        {
          "confidence": 0.0939941480755806,
          "label": "art"
        },
        {
          "confidence": 0.0939941480755806,
          "label": "decoration"
        },
        {
          "confidence": 0.093994140625,
          "label": "jewelry"
        },
        {
          "confidence": 0.028564453125,
          "label": "military_uniform"
        },
        {
          "confidence": 0.022216796875,
          "label": "bowtie"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.31146812438964844,
          "width": 0.4231586456298828,
          "x": 0.22657814621925354,
          "y": 0.4245191812515259
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Kiss - Gustav Klimt, Vienna 1908
classify 1908 Public Domain (Klimt died 1918)

The Kiss - Gustav Klimt, Vienna 1908

Klimt's gold-leaf masterpiece in the Belvedere, Vienna. A genuine challenge for Vision: heavy ornament, abstract gold, two faces partially visible.

$ auge --all klimt-kiss.jpg
A OCR 2 lines · 9 chars
COS
Tot g
# Classify 10 labels
people
74%
adult
74%
clothing
33%
jeans
33%
structure
26%
wood_processed
26%
art
22%
painting
19%
machine
18%
consumer_electronics
18%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/klimt-kiss.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7396463751792908,
          "label": "people"
        },
        {
          "confidence": 0.7392578125,
          "label": "adult"
        },
        {
          "confidence": 0.3319745659828186,
          "label": "clothing"
        },
        {
          "confidence": 0.327392578125,
          "label": "jeans"
        },
        {
          "confidence": 0.25930631160736084,
          "label": "structure"
        },
        {
          "confidence": 0.25927734375,
          "label": "wood_processed"
        },
        {
          "confidence": 0.2198285013437271,
          "label": "art"
        },
        {
          "confidence": 0.189453125,
          "label": "painting"
        },
        {
          "confidence": 0.18021178245544434,
          "label": "machine"
        },
        {
          "confidence": 0.1802116483449936,
          "label": "consumer_electronics"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "COS",
        "Tot g"
      ],
      "text": "COS\nTot g"
    }
  }
}
Wikimedia source
EAN-13 supermarket barcode
barcode 2026 CC0 (locally generated, real ISBN-13 of The Mythical Man-Month)

EAN-13 supermarket barcode

The barcode you see on every supermarket product. EAN-13 (European Article Number) is the global standard. This one encodes ISBN-13 9780201379624 - The Mythical Man-Month.

$ auge --all ean13-supermarket.png
A OCR 3 lines · 15 chars
9
780201
379624
# Classify 10 labels
document
8%
art
8%
illustrations
8%
printed_page
8%
chart
4%
diagram
4%
material
3%
handwriting
2%
structure
2%
fence
2%
Barcodes 1 found
EAN13 9780201379624
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/ean13-supermarket.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "9780201379624",
          "symbology": "EAN13"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.08468558639287949,
          "label": "document"
        },
        {
          "confidence": 0.0839843824505806,
          "label": "art"
        },
        {
          "confidence": 0.083984375,
          "label": "illustrations"
        },
        {
          "confidence": 0.083251953125,
          "label": "printed_page"
        },
        {
          "confidence": 0.04248046875,
          "label": "chart"
        },
        {
          "confidence": 0.04248046875,
          "label": "diagram"
        },
        {
          "confidence": 0.032227661460638046,
          "label": "material"
        },
        {
          "confidence": 0.019287109375,
          "label": "handwriting"
        },
        {
          "confidence": 0.01803063414990902,
          "label": "structure"
        },
        {
          "confidence": 0.017822265625,
          "label": "fence"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "9",
        "780201",
        "379624"
      ],
      "text": "9\n780201\n379624"
    }
  }
}
Wikimedia source
Two women operating ENIAC, 1946
faces 1946 Public Domain (US Army photo)

Two women operating ENIAC, 1946

Frances Bilas and Betty Jean Jennings - two of the six women who programmed ENIAC, the first general-purpose digital electronic computer, by physically rewiring it. The original software engineers.

$ auge --all eniac-1946.jpg
A OCR 1 line · 9 chars
BEE ESIMI
# Classify 10 labels
people
70%
adult
70%
clothing
24%
container
17%
luggage
17%
purse
17%
structure
14%
sign
13%
suit
12%
bag
8%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/eniac-1946.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7045940160751343,
          "label": "people"
        },
        {
          "confidence": 0.70458984375,
          "label": "adult"
        },
        {
          "confidence": 0.23757322132587433,
          "label": "clothing"
        },
        {
          "confidence": 0.17425985634326935,
          "label": "container"
        },
        {
          "confidence": 0.17425885796546936,
          "label": "luggage"
        },
        {
          "confidence": 0.17236328125,
          "label": "purse"
        },
        {
          "confidence": 0.13719545304775238,
          "label": "structure"
        },
        {
          "confidence": 0.13306382298469543,
          "label": "sign"
        },
        {
          "confidence": 0.117431640625,
          "label": "suit"
        },
        {
          "confidence": 0.07812614738941193,
          "label": "bag"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "BEE ESIMI"
      ],
      "text": "BEE ESIMI"
    }
  }
}
Wikimedia source
Hubble Ultra-Deep Field, 2004
classify 2004 Public Domain (NASA / ESA)

Hubble Ultra-Deep Field, 2004

Eleven days of exposure in a single patch of sky one-tenth the diameter of a full moon. Roughly 10,000 galaxies, the most distant from when the universe was 800 million years old.

$ auge --all hubble-deep-field.jpg
A OCR no text
No text detected in this image.
# Classify 6 labels
night_sky
87%
outdoor
87%
sky
87%
fire
2%
pyrotechnics
2%
fireworks
2%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/hubble-deep-field.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8671875,
          "label": "night_sky"
        },
        {
          "confidence": 0.8671875,
          "label": "outdoor"
        },
        {
          "confidence": 0.8671875,
          "label": "sky"
        },
        {
          "confidence": 0.02270507998764515,
          "label": "fire"
        },
        {
          "confidence": 0.02270507998764515,
          "label": "pyrotechnics"
        },
        {
          "confidence": 0.022705078125,
          "label": "fireworks"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Goethe in the Roman Campagna - Tischbein, 1787
face 1
historical 1787 Public Domain

Goethe in the Roman Campagna - Tischbein, 1787

The icon of Weimar Classicism. Tischbein painted his friend Goethe reclining among Roman ruins - face detection on a 240-year-old oil painting.

$ auge --all goethe-campagna.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
84%
painting
84%
people
43%
adult
43%
outdoor
32%
sky
32%
cloudy
32%
plant
31%
foliage
31%
blue_sky
13%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.217 y=0.662 w=0.070 h=0.089
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/goethe-campagna.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8417993187904358,
          "label": "art"
        },
        {
          "confidence": 0.841796875,
          "label": "painting"
        },
        {
          "confidence": 0.4328780770301819,
          "label": "people"
        },
        {
          "confidence": 0.432861328125,
          "label": "adult"
        },
        {
          "confidence": 0.31955671310424805,
          "label": "outdoor"
        },
        {
          "confidence": 0.31845083832740784,
          "label": "sky"
        },
        {
          "confidence": 0.316162109375,
          "label": "cloudy"
        },
        {
          "confidence": 0.30737563967704773,
          "label": "plant"
        },
        {
          "confidence": 0.307373046875,
          "label": "foliage"
        },
        {
          "confidence": 0.131103515625,
          "label": "blue_sky"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.08913090080022812,
          "width": 0.07005131989717484,
          "x": 0.21706023812294006,
          "y": 0.6615588665008545
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Bird of Washington - John James Audubon, 1838
classify 1838 Public Domain

Bird of Washington - John James Audubon, 1838

From Audubon's Birds of America - the most expensive printed book in the world (a complete copy sold for $11.5M in 2010). Hand-coloured engraving of a bird that may not have existed.

$ auge --all audubon-bird.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
64%
painting
59%
illustrations
48%
animal
25%
bird
25%
poultry
25%
raptor
15%
vulture
15%
mammal
3%
eagle
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/audubon-bird.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6444507241249084,
          "label": "art"
        },
        {
          "confidence": 0.59423828125,
          "label": "painting"
        },
        {
          "confidence": 0.4833984375,
          "label": "illustrations"
        },
        {
          "confidence": 0.25273245573043823,
          "label": "animal"
        },
        {
          "confidence": 0.2527138888835907,
          "label": "bird"
        },
        {
          "confidence": 0.245361328125,
          "label": "poultry"
        },
        {
          "confidence": 0.14611941576004028,
          "label": "raptor"
        },
        {
          "confidence": 0.14599609375,
          "label": "vulture"
        },
        {
          "confidence": 0.033114299178123474,
          "label": "mammal"
        },
        {
          "confidence": 0.032470703125,
          "label": "eagle"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Hunters in the Snow - Pieter Bruegel, 1565
classify 1565 Public Domain

Hunters in the Snow - Pieter Bruegel, 1565

Bruegel's January from his Months series. A complex Flemish winter scene - perfect for stress-testing classification (snow, hunters, dogs, village, ice skating).

$ auge --all bruegel-hunters.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
liquid
74%
water
74%
frozen
74%
snow
74%
art
32%
painting
32%
plant
30%
branch
30%
people
29%
adult
29%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/bruegel-hunters.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7426769733428955,
          "label": "liquid"
        },
        {
          "confidence": 0.7426769733428955,
          "label": "water"
        },
        {
          "confidence": 0.74267578125,
          "label": "frozen"
        },
        {
          "confidence": 0.74267578125,
          "label": "snow"
        },
        {
          "confidence": 0.31983035802841187,
          "label": "art"
        },
        {
          "confidence": 0.31982421875,
          "label": "painting"
        },
        {
          "confidence": 0.29809704422950745,
          "label": "plant"
        },
        {
          "confidence": 0.298095703125,
          "label": "branch"
        },
        {
          "confidence": 0.2925792634487152,
          "label": "people"
        },
        {
          "confidence": 0.29248046875,
          "label": "adult"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Mozart, by Lange (unfinished, 1782)
face 1
faces 1782 Public Domain

Mozart, by Lange (unfinished, 1782)

Joseph Lange's unfinished oil portrait of his brother-in-law - widely considered the most accurate likeness of Mozart. auge detects the painted face.

$ auge --all mozart.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
56%
painting
56%
people
50%
adult
50%
illustrations
4%
clothing
2%
suit
1%
child
1%
military_uniform
1%
tool
1%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.180 y=0.324 w=0.407 h=0.294
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/mozart.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.5634788274765015,
          "label": "art"
        },
        {
          "confidence": 0.5634765625,
          "label": "painting"
        },
        {
          "confidence": 0.5034710168838501,
          "label": "people"
        },
        {
          "confidence": 0.49609375,
          "label": "adult"
        },
        {
          "confidence": 0.03662109375,
          "label": "illustrations"
        },
        {
          "confidence": 0.015756629407405853,
          "label": "clothing"
        },
        {
          "confidence": 0.013916015625,
          "label": "suit"
        },
        {
          "confidence": 0.01318359375,
          "label": "child"
        },
        {
          "confidence": 0.01220703125,
          "label": "military_uniform"
        },
        {
          "confidence": 0.010728261433541775,
          "label": "tool"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.29436373710632324,
          "width": 0.4068198800086975,
          "x": 0.18011009693145752,
          "y": 0.32432639598846436
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Bilingual French-Dutch street sign, Brussels
multilingual 2008 Public Domain

Bilingual French-Dutch street sign, Brussels

Brussels has been officially bilingual since 1932. auge --langs fr-FR,nl-NL handles both scripts on the same sign in one pass.

$ auge --all --langs fr-FR,nl-NL brussels-bilingual.jpg
A OCR 8 lines · 77 chars
VILLE DE BRUXELLES
STAD
BRUSSEL
PLACE
ROYALE
KONINGS
PLEIN
POLICE - POLITIE -
# Classify 10 labels
structure
58%
sign
58%
street_sign
58%
art
10%
decoration
10%
container
6%
cardboard_box
6%
wood_processed
5%
material
4%
brick
4%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/brussels-bilingual.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.5805936455726624,
          "label": "structure"
        },
        {
          "confidence": 0.5805856585502625,
          "label": "sign"
        },
        {
          "confidence": 0.58056640625,
          "label": "street_sign"
        },
        {
          "confidence": 0.10023181885480881,
          "label": "art"
        },
        {
          "confidence": 0.1000976637005806,
          "label": "decoration"
        },
        {
          "confidence": 0.05663655698299408,
          "label": "container"
        },
        {
          "confidence": 0.056396484375,
          "label": "cardboard_box"
        },
        {
          "confidence": 0.051513671875,
          "label": "wood_processed"
        },
        {
          "confidence": 0.042266085743904114,
          "label": "material"
        },
        {
          "confidence": 0.042236328125,
          "label": "brick"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "VILLE DE BRUXELLES",
        "STAD",
        "BRUSSEL",
        "PLACE",
        "ROYALE",
        "KONINGS",
        "PLEIN",
        "POLICE - POLITIE -"
      ],
      "text": "VILLE DE BRUXELLES\nSTAD\nBRUSSEL\nPLACE\nROYALE\nKONINGS\nPLEIN\nPOLICE - POLITIE -"
    }
  }
}
Wikimedia source
QR code - vCard contact
barcode 2026 CC0 (locally generated)

QR code - vCard contact

A vCard 3.0 contact card encoded in a QR. auge decodes the multi-line text and you can save it as a contact - all without touching the cloud.

$ auge --all qr-vcard.png
A OCR no text
No text detected in this image.
# Classify 10 labels
document
5%
printed_page
5%
art
5%
illustrations
5%
structure
3%
sign
3%
screenshot
2%
decoration
2%
frame
2%
material
2%
Barcodes 1 found
QR BEGIN:VCARD VERSION:3.0 FN:Arthur Ficial ORG:Apfel ecosystem EMAIL:arti.ficial@fullstackoptimization.com URL:https://auge.franzai.com END:VCARD
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-vcard.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "BEGIN:VCARD\nVERSION:3.0\nFN:Arthur Ficial\nORG:Apfel ecosystem\nEMAIL:arti.ficial@fullstackoptimization.com\nURL:https://auge.franzai.com\nEND:VCARD",
          "symbology": "QR"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.05432205647230148,
          "label": "document"
        },
        {
          "confidence": 0.0537109375,
          "label": "printed_page"
        },
        {
          "confidence": 0.051669299602508545,
          "label": "art"
        },
        {
          "confidence": 0.051513671875,
          "label": "illustrations"
        },
        {
          "confidence": 0.03345812112092972,
          "label": "structure"
        },
        {
          "confidence": 0.03344736993312836,
          "label": "sign"
        },
        {
          "confidence": 0.024658203125,
          "label": "screenshot"
        },
        {
          "confidence": 0.01708986423909664,
          "label": "decoration"
        },
        {
          "confidence": 0.01708984375,
          "label": "frame"
        },
        {
          "confidence": 0.016856322064995766,
          "label": "material"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Eiffel Tower, Paris
classify 2010 CC0

Eiffel Tower, Paris

1,083 feet of wrought iron. Classify identifies it - and Vision is good enough to recognise specific landmarks in its training set.

$ auge --all eiffel-tower.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
outdoor
92%
sky
86%
blue_sky
86%
land
77%
grass
77%
structure
70%
arch
70%
fence
6%
dome
5%
path
5%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/eiffel-tower.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.916789710521698,
          "label": "outdoor"
        },
        {
          "confidence": 0.8598634004592896,
          "label": "sky"
        },
        {
          "confidence": 0.85986328125,
          "label": "blue_sky"
        },
        {
          "confidence": 0.7705096006393433,
          "label": "land"
        },
        {
          "confidence": 0.7705078125,
          "label": "grass"
        },
        {
          "confidence": 0.7037513852119446,
          "label": "structure"
        },
        {
          "confidence": 0.70361328125,
          "label": "arch"
        },
        {
          "confidence": 0.057373046875,
          "label": "fence"
        },
        {
          "confidence": 0.0537109375,
          "label": "dome"
        },
        {
          "confidence": 0.04736328125,
          "label": "path"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Beethoven with the Missa solemnis manuscript
face 1
faces 1820 Public Domain

Beethoven with the Missa solemnis manuscript

Stieler's 1820 portrait. Possible to OCR the manuscript score in his hand? auge runs every analysis to find out.

$ auge --all beethoven.jpg
A OCR 2 lines · 15 chars
shiple punis
**
# Classify 10 labels
people
44%
adult
44%
art
31%
painting
31%
document
17%
book
17%
clothing
14%
cloak
14%
illustrations
10%
suit
7%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.365 y=0.536 w=0.302 h=0.243
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/beethoven.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.44067397713661194,
          "label": "people"
        },
        {
          "confidence": 0.440673828125,
          "label": "adult"
        },
        {
          "confidence": 0.3063300549983978,
          "label": "art"
        },
        {
          "confidence": 0.305419921875,
          "label": "painting"
        },
        {
          "confidence": 0.16631679236888885,
          "label": "document"
        },
        {
          "confidence": 0.166259765625,
          "label": "book"
        },
        {
          "confidence": 0.1412958800792694,
          "label": "clothing"
        },
        {
          "confidence": 0.137939453125,
          "label": "cloak"
        },
        {
          "confidence": 0.1005859375,
          "label": "illustrations"
        },
        {
          "confidence": 0.0703125,
          "label": "suit"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.24293918907642365,
          "width": 0.3023454248905182,
          "x": 0.36529967188835144,
          "y": 0.5363717675209045
        }
      ]
    },
    "ocr": {
      "lines": [
        "shiple punis",
        "**"
      ],
      "text": "shiple punis\n**"
    }
  }
}
Wikimedia source
Trilingual sign in Jaffa: Arabic, Hebrew, English
multilingual 2009 Public Domain

Trilingual sign in Jaffa: Arabic, Hebrew, English

Three scripts (Latin, Arabic abjad, Hebrew abjad) on a single street sign in Jaffa. A serious test of right-to-left + multi-script OCR.

$ auge --all --langs en-US,ar,he jaffa-trilingual.jpg
A OCR 24 lines · 727 chars
157
The Gate to the Ancient World
The Jaffa port is one of the oldest
ports in the world, in almost
continuous operation for the
past 4,000 years. The port takes
advantage of calcareous sandstone
(kurkar) reefs that create a natural
breakwater and safe haven.
Beginning in the mid- 19th century the
port was expanded and upgraded,
but it still could not accommodate
large steamships. These anchored at
sea, with cargo and passengers
ferried to land in small boats.
The British Mandate saw construction
of a concrete pier and breakwater
along with large warehouses and
cranes. During the 20th century the
port gradually began to sink; after 1965
it was no longer used for cargo.
The Jaffa Port
→ t in de 4000 bao
1965 pk + 3,4
# Classify 10 labels
structure
36%
rocks
36%
art
30%
decoration
30%
material
28%
brick
28%
plant
19%
foliage
18%
sign
18%
fence
11%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/jaffa-trilingual.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.3624260127544403,
          "label": "structure"
        },
        {
          "confidence": 0.3564453125,
          "label": "rocks"
        },
        {
          "confidence": 0.29568323493003845,
          "label": "art"
        },
        {
          "confidence": 0.2956830561161041,
          "label": "decoration"
        },
        {
          "confidence": 0.282728374004364,
          "label": "material"
        },
        {
          "confidence": 0.28271484375,
          "label": "brick"
        },
        {
          "confidence": 0.19067680835723877,
          "label": "plant"
        },
        {
          "confidence": 0.184814453125,
          "label": "foliage"
        },
        {
          "confidence": 0.17627137899398804,
          "label": "sign"
        },
        {
          "confidence": 0.11279296875,
          "label": "fence"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "157",
        "The Gate to the Ancient World",
        "The Jaffa port is one of the oldest",
        "ports in the world, in almost",
        "continuous operation for the",
        "past 4,000 years. The port takes",
        "advantage of calcareous sandstone",
        "(kurkar) reefs that create a natural",
        "breakwater and safe haven.",
        "Beginning in the mid- 19th century the",
        "port was expanded and upgraded,",
        "but it still could not accommodate",
        "large steamships. These anchored at",
        "sea, with cargo and passengers",
        "ferried to land in small boats.",
        "The British Mandate saw construction",
        "of a concrete pier and breakwater",
        "along with large warehouses and",
        "cranes. During the 20th century the",
        "port gradually began to sink; after 1965",
        "it was no longer used for cargo.",
        "The Jaffa Port",
        "→ t in de 4000 bao",
        "1965 pk + 3,4"
      ],
      "text": "157\nThe Gate to the Ancient World\nThe Jaffa port is one of the oldest\nports in the world, in almost\ncontinuous operation for the\npast 4,000 years. The port takes\nadvantage of calcareous sandstone\n(kurkar) reefs that create a natural\nbreakwater and safe haven.\nBeginning in the mid- 19th century the\nport was expanded and upgraded,\nbut it still could not accommodate\nlarge steamships. These anchored at\nsea, with cargo and passengers\nferried to land in small boats.\nThe British Mandate saw construction\nof a concrete pier and breakwater\nalong with large warehouses and\ncranes. During the 20th century the\nport gradually began to sink; after 1965\nit was no longer used for cargo.\nThe Jaffa Port\n→ t in de 4000 bao\n1965 pk + 3,4"
    }
  }
}
Wikimedia source
Tournée du Chat Noir - Steinlen, 1896
face 1
multilingual 1896 Public Domain (Steinlen died 1923)

Tournée du Chat Noir - Steinlen, 1896

Steinlen's Belle-Époque poster of a giant black cat over Montmartre. A founding piece of poster art and one of the most-reproduced cat images in history. OCRs the French title.

$ auge --all --langs fr-FR steinlen-cat.jpg
A OCR 6 lines · 92 chars
cLinioue
CheRon
MÉDECİN VETERINAIRE SPÉCIALISTE
8RUE des MOULINS PARİS
ANATORIUM
PENSION
# Classify 10 labels
art
78%
illustrations
78%
sign
19%
structure
19%
document
12%
printed_page
12%
people
7%
adult
7%
animal
5%
mammal
5%
Barcodes none
No barcodes or QR codes detected.
Faces 1 detected
1 face located. Bounding boxes overlaid on the image above.
face 1: x=0.326 y=0.808 w=0.108 h=0.073
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/steinlen-cat.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7836915254592896,
          "label": "art"
        },
        {
          "confidence": 0.78369140625,
          "label": "illustrations"
        },
        {
          "confidence": 0.191650390625,
          "label": "sign"
        },
        {
          "confidence": 0.191650390625,
          "label": "structure"
        },
        {
          "confidence": 0.11744113266468048,
          "label": "document"
        },
        {
          "confidence": 0.117431640625,
          "label": "printed_page"
        },
        {
          "confidence": 0.07202398777008057,
          "label": "people"
        },
        {
          "confidence": 0.072021484375,
          "label": "adult"
        },
        {
          "confidence": 0.05413498356938362,
          "label": "animal"
        },
        {
          "confidence": 0.05413464829325676,
          "label": "mammal"
        }
      ]
    },
    "faces": {
      "count": 1,
      "faces": [
        {
          "height": 0.07319926470518112,
          "width": 0.10755810141563416,
          "x": 0.3258713185787201,
          "y": 0.8082327246665955
        }
      ]
    },
    "ocr": {
      "lines": [
        "cLinioue",
        "CheRon",
        "MÉDECİN VETERINAIRE SPÉCIALISTE",
        "8RUE des MOULINS PARİS",
        "ANATORIUM",
        "PENSION"
      ],
      "text": "cLinioue\nCheRon\nMÉDECİN VETERINAIRE SPÉCIALISTE\n8RUE des MOULINS PARİS\nANATORIUM\nPENSION"
    }
  }
}
Wikimedia source
The Ambassadors - Hans Holbein, 1533
face 1
face 2
faces 1533 Public Domain

The Ambassadors - Hans Holbein, 1533

Two French ambassadors at the court of Henry VIII, surrounded by symbols of learning - and the famous anamorphic skull. Two faces, dense iconography.

$ auge --all holbein-ambassadors.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
people
73%
adult
73%
structure
52%
wood_processed
52%
material
49%
textile
49%
celebration
22%
costume
22%
furniture
18%
table
17%
Barcodes none
No barcodes or QR codes detected.
Faces 2 detected
2 faces located. Bounding boxes overlaid on the image above.
face 1: x=0.210 y=0.786 w=0.082 h=0.083
face 2: x=0.804 y=0.775 w=0.078 h=0.079
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/holbein-ambassadors.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7304688096046448,
          "label": "people"
        },
        {
          "confidence": 0.73046875,
          "label": "adult"
        },
        {
          "confidence": 0.52397620677948,
          "label": "structure"
        },
        {
          "confidence": 0.5224609375,
          "label": "wood_processed"
        },
        {
          "confidence": 0.48658713698387146,
          "label": "material"
        },
        {
          "confidence": 0.485595703125,
          "label": "textile"
        },
        {
          "confidence": 0.22022099792957306,
          "label": "celebration"
        },
        {
          "confidence": 0.22021484375,
          "label": "costume"
        },
        {
          "confidence": 0.17500130832195282,
          "label": "furniture"
        },
        {
          "confidence": 0.174560546875,
          "label": "table"
        }
      ]
    },
    "faces": {
      "count": 2,
      "faces": [
        {
          "height": 0.08345133066177368,
          "width": 0.08221259713172913,
          "x": 0.20964355766773224,
          "y": 0.7861505746841431
        },
        {
          "height": 0.07872382551431656,
          "width": 0.07755526900291443,
          "x": 0.8036110997200012,
          "y": 0.7754983901977539
        }
      ]
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Sterzing / Vipiteno - South Tyrol bilingual sign
multilingual 2014 Public Domain

Sterzing / Vipiteno - South Tyrol bilingual sign

South Tyrol road sign: every place name in German and Italian. Sterzing in German, Vipiteno in Italian, same town. auge --langs de-DE,it-IT.

$ auge --all --langs de-DE,it-IT south-tyrol-sign.jpg
A OCR 5 lines · 41 chars
STERZING
VIPITENO
Willkommen
Benvenuti
54
# Classify 10 labels
sign
25%
structure
25%
street_sign
25%
document
5%
screenshot
5%
outdoor
3%
sky
3%
blue_sky
3%
art
2%
illustrations
2%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/south-tyrol-sign.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.25276443362236023,
          "label": "sign"
        },
        {
          "confidence": 0.25276443362236023,
          "label": "structure"
        },
        {
          "confidence": 0.252685546875,
          "label": "street_sign"
        },
        {
          "confidence": 0.04839390888810158,
          "label": "document"
        },
        {
          "confidence": 0.0458984375,
          "label": "screenshot"
        },
        {
          "confidence": 0.03155980631709099,
          "label": "outdoor"
        },
        {
          "confidence": 0.03155971318483353,
          "label": "sky"
        },
        {
          "confidence": 0.031494140625,
          "label": "blue_sky"
        },
        {
          "confidence": 0.021487830206751823,
          "label": "art"
        },
        {
          "confidence": 0.021484375,
          "label": "illustrations"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "STERZING",
        "VIPITENO",
        "Willkommen",
        "Benvenuti",
        "54"
      ],
      "text": "STERZING\nVIPITENO\nWillkommen\nBenvenuti\n54"
    }
  }
}
Wikimedia source
Anatomy of the Horse - George Stubbs, 1766
classify 1766 Public Domain

Anatomy of the Horse - George Stubbs, 1766

George Stubbs spent 18 months dissecting horses to draw them perfectly. The resulting 1766 atlas is half scientific document, half love letter to the animal.

$ auge --all stubbs-horse.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
62%
illustrations
62%
animal
51%
mammal
51%
ungulates
51%
horse
51%
feline
4%
lion
4%
camel
4%
canine
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/stubbs-horse.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6176758408546448,
          "label": "art"
        },
        {
          "confidence": 0.61767578125,
          "label": "illustrations"
        },
        {
          "confidence": 0.5112391710281372,
          "label": "animal"
        },
        {
          "confidence": 0.5112391710281372,
          "label": "mammal"
        },
        {
          "confidence": 0.5112342238426208,
          "label": "ungulates"
        },
        {
          "confidence": 0.51123046875,
          "label": "horse"
        },
        {
          "confidence": 0.03590106964111328,
          "label": "feline"
        },
        {
          "confidence": 0.035888671875,
          "label": "lion"
        },
        {
          "confidence": 0.03515625,
          "label": "camel"
        },
        {
          "confidence": 0.03279639035463333,
          "label": "canine"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
UPC-A grocery barcode
barcode 2026 CC0 (locally generated)

UPC-A grocery barcode

UPC-A is the 12-digit barcode used on most US groceries. auge --barcode reports it normalised to EAN-13 with a leading zero - same number, same on-device decoder.

$ auge --all upca-grocery.png
A OCR 2 lines · 13 chars
12345 67890
5
# Classify 10 labels
art
9%
illustrations
9%
document
7%
chart
6%
diagram
6%
printed_page
6%
handwriting
2%
material
2%
structure
1%
fence
1%
Barcodes 1 found
EAN13 0012345678905
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/upca-grocery.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "0012345678905",
          "symbology": "EAN13"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.093505859375,
          "label": "art"
        },
        {
          "confidence": 0.093505859375,
          "label": "illustrations"
        },
        {
          "confidence": 0.07124409824609756,
          "label": "document"
        },
        {
          "confidence": 0.060546875,
          "label": "chart"
        },
        {
          "confidence": 0.060546875,
          "label": "diagram"
        },
        {
          "confidence": 0.05908203125,
          "label": "printed_page"
        },
        {
          "confidence": 0.019287109375,
          "label": "handwriting"
        },
        {
          "confidence": 0.017096085473895073,
          "label": "material"
        },
        {
          "confidence": 0.011113823391497135,
          "label": "structure"
        },
        {
          "confidence": 0.010498046875,
          "label": "fence"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "12345 67890",
        "5"
      ],
      "text": "12345 67890\n5"
    }
  }
}
Wikimedia source
Vienna State Opera House
classify 2006 CC0

Vienna State Opera House

The Wiener Staatsoper, opened 1869. Neo-Renaissance architecture, classified by auge with crisp confidence.

$ auge --all vienna-opera.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
outdoor
98%
land
95%
road
94%
street
94%
sky
89%
cloudy
89%
machine
77%
vehicle
77%
automobile
77%
car
77%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/vienna-opera.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.9811716079711914,
          "label": "outdoor"
        },
        {
          "confidence": 0.9458186626434326,
          "label": "land"
        },
        {
          "confidence": 0.9384765625,
          "label": "road"
        },
        {
          "confidence": 0.9384765625,
          "label": "street"
        },
        {
          "confidence": 0.8921303153038025,
          "label": "sky"
        },
        {
          "confidence": 0.89208984375,
          "label": "cloudy"
        },
        {
          "confidence": 0.7748026251792908,
          "label": "machine"
        },
        {
          "confidence": 0.7748026251792908,
          "label": "vehicle"
        },
        {
          "confidence": 0.7748003602027893,
          "label": "automobile"
        },
        {
          "confidence": 0.7744140625,
          "label": "car"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
QR code on Conwy street info sign (UK, outdoor)
barcode 2012 CC BY-SA 2.0 (Geograph)

QR code on Conwy street info sign (UK, outdoor)

A real-world outdoor QR code on a tourist information sign in Conwy, Wales. auge decodes the URL plus picks up the surrounding sign text.

$ auge --all qr-outdoor-conwy.jpg
A OCR 9 lines · 119 chars
BROKAP
EXCE
Paentiad JC Ibbetson 1794
JC Ibbetson painting 1794
Discover history
n
CONWY
historypoints.org
More history
# Classify 10 labels
outdoor
89%
sky
89%
cloudy
79%
blue_sky
78%
structure
34%
mast
32%
pole
32%
machine
23%
vehicle
23%
watercraft
23%
Barcodes 2 found
QR JC Ibbetson painted the castle and foreshore in moonlight from this direction c.1794. To compare his image with today’s scene, scan the lower QR barcode
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-outdoor-conwy.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "JC Ibbetson painted the castle and foreshore in moonlight from this direction c.1794. To compare his image with today’s scene, scan the lower QR barcode ",
          "symbology": "QR"
        },
        {
          "payload": "http://historypoints.org/index.php?page=jc-ibbetson-painting-of-1794",
          "symbology": "QR"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8883610963821411,
          "label": "outdoor"
        },
        {
          "confidence": 0.8883607387542725,
          "label": "sky"
        },
        {
          "confidence": 0.7919921875,
          "label": "cloudy"
        },
        {
          "confidence": 0.7841796875,
          "label": "blue_sky"
        },
        {
          "confidence": 0.33996233344078064,
          "label": "structure"
        },
        {
          "confidence": 0.32080078125,
          "label": "mast"
        },
        {
          "confidence": 0.32080078125,
          "label": "pole"
        },
        {
          "confidence": 0.22828108072280884,
          "label": "machine"
        },
        {
          "confidence": 0.2282789647579193,
          "label": "vehicle"
        },
        {
          "confidence": 0.22827866673469543,
          "label": "watercraft"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "BROKAP",
        "EXCE",
        "Paentiad JC Ibbetson 1794",
        "JC Ibbetson painting 1794",
        "Discover history",
        "n",
        "CONWY",
        "historypoints.org",
        "More history"
      ],
      "text": "BROKAP\nEXCE\nPaentiad JC Ibbetson 1794\nJC Ibbetson painting 1794\nDiscover history\nn\nCONWY\nhistorypoints.org\nMore history"
    }
  }
}
Wikimedia source
Colosseum, Rome
classify 2007 CC0

Colosseum, Rome

The Flavian Amphitheatre, opened 80 AD. Classify nails 'amphitheatre / ruin / arena' - and auge probably detects faces in the tourist crowd.

$ auge --all colosseum.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
structure
76%
arch
75%
outdoor
68%
land
61%
grass
60%
sky
54%
blue_sky
49%
cloudy
41%
fence
30%
path
29%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/colosseum.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7610287666320801,
          "label": "structure"
        },
        {
          "confidence": 0.7529296875,
          "label": "arch"
        },
        {
          "confidence": 0.6766185164451599,
          "label": "outdoor"
        },
        {
          "confidence": 0.6088550090789795,
          "label": "land"
        },
        {
          "confidence": 0.6015625,
          "label": "grass"
        },
        {
          "confidence": 0.5377476215362549,
          "label": "sky"
        },
        {
          "confidence": 0.488037109375,
          "label": "blue_sky"
        },
        {
          "confidence": 0.410888671875,
          "label": "cloudy"
        },
        {
          "confidence": 0.304931640625,
          "label": "fence"
        },
        {
          "confidence": 0.29370129108428955,
          "label": "path"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Notre-Dame de Paris, west façade
classify 2014 CC0

Notre-Dame de Paris, west façade

The Gothic west façade, before the 2019 fire. Sculpture-dense - auge looks for faces in the carved statues alongside cathedral classification.

$ auge --all notre-dame.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
structure
56%
arch
56%
art
17%
fence
15%
material
12%
brick
12%
dome
10%
conveyance
7%
portal
7%
window
6%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/notre-dame.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.5623368620872498,
          "label": "structure"
        },
        {
          "confidence": 0.560546875,
          "label": "arch"
        },
        {
          "confidence": 0.1682402640581131,
          "label": "art"
        },
        {
          "confidence": 0.15185546875,
          "label": "fence"
        },
        {
          "confidence": 0.11792084574699402,
          "label": "material"
        },
        {
          "confidence": 0.117919921875,
          "label": "brick"
        },
        {
          "confidence": 0.096923828125,
          "label": "dome"
        },
        {
          "confidence": 0.06773553788661957,
          "label": "conveyance"
        },
        {
          "confidence": 0.06772664189338684,
          "label": "portal"
        },
        {
          "confidence": 0.057861328125,
          "label": "window"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Dodo - George Edwards, 1759
classify 1759 Public Domain

The Dodo - George Edwards, 1759

George Edwards's 1759 illustration of a Dodo - an animal already extinct for nearly a century when he drew it, from a Dutch painting. The most charming dead bird in art history.

$ auge --all edwards-dodo.jpg
A OCR 3 lines · 42 chars
the DoDo & Given
by GEDWARDS
FRS. AD. J59.
# Classify 10 labels
art
74%
painting
74%
animal
28%
bird
28%
poultry
28%
illustrations
21%
outdoor
11%
grass
11%
land
11%
plant
9%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/edwards-dodo.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.740666925907135,
          "label": "art"
        },
        {
          "confidence": 0.73974609375,
          "label": "painting"
        },
        {
          "confidence": 0.2799437642097473,
          "label": "animal"
        },
        {
          "confidence": 0.2799203395843506,
          "label": "bird"
        },
        {
          "confidence": 0.279541015625,
          "label": "poultry"
        },
        {
          "confidence": 0.214599609375,
          "label": "illustrations"
        },
        {
          "confidence": 0.11426381021738052,
          "label": "outdoor"
        },
        {
          "confidence": 0.11376953125,
          "label": "grass"
        },
        {
          "confidence": 0.11376953125,
          "label": "land"
        },
        {
          "confidence": 0.09080713242292404,
          "label": "plant"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "the DoDo & Given",
        "by GEDWARDS",
        "FRS. AD. J59."
      ],
      "text": "the DoDo & Given\nby GEDWARDS\nFRS. AD. J59."
    }
  }
}
Wikimedia source
Belgrade street sign - Cyrillic and Latin
multilingual 2009 Public Domain

Belgrade street sign - Cyrillic and Latin

Serbia uses both Cyrillic and Latin scripts officially. This Belgrade plate names George Washington in both alphabets. auge OCRs sr-Cyrl + sr-Latn at the same time.

$ auge --all --langs sr-Cyrl,sr-Latn,en-US belgrade-cyrillic.jpg
A OCR 6 lines · 70 chars
УЛИЦА
ЧОРЦА
ВАШИНГТОНА
ULICA
DŽORDŽA
VAŠINGTONA
# Classify 10 labels
art
36%
decoration
36%
structure
20%
chalkboard
19%
sign
11%
wood_processed
7%
frame
6%
illustrations
5%
document
3%
handwriting
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/belgrade-cyrillic.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.3594709634780884,
          "label": "art"
        },
        {
          "confidence": 0.3594427704811096,
          "label": "decoration"
        },
        {
          "confidence": 0.19719330966472626,
          "label": "structure"
        },
        {
          "confidence": 0.19140625,
          "label": "chalkboard"
        },
        {
          "confidence": 0.10931399464607239,
          "label": "sign"
        },
        {
          "confidence": 0.072265625,
          "label": "wood_processed"
        },
        {
          "confidence": 0.059814453125,
          "label": "frame"
        },
        {
          "confidence": 0.0478515625,
          "label": "illustrations"
        },
        {
          "confidence": 0.03433666005730629,
          "label": "document"
        },
        {
          "confidence": 0.032958984375,
          "label": "handwriting"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "УЛИЦА",
        "ЧОРЦА",
        "ВАШИНГТОНА",
        "ULICA",
        "DŽORDŽA",
        "VAŠINGTONA"
      ],
      "text": "УЛИЦА\nЧОРЦА\nВАШИНГТОНА\nULICA\nDŽORDŽA\nVAŠINGTONA"
    }
  }
}
Wikimedia source
DataMatrix code on a bottle of milk
barcode 2018 CC0

DataMatrix code on a bottle of milk

A real DataMatrix barcode in the wild - on a milk bottle. Vision recognises this format alongside QR, EAN, Code128, PDF417, and Aztec.

$ auge --all datamatrix-milk.jpg
A OCR 1 line · 2 chars
26
# Classify 10 labels
machine
12%
keypad
12%
tableware
6%
utensil
6%
plate
6%
structure
3%
sign
3%
material
2%
textile
2%
document
1%
Barcodes 1 found
DataMatrix 0104607078117294215fGHNL93H/xN
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/datamatrix-milk.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "0104607078117294215fGHNL\u001d93H/xN",
          "symbology": "DataMatrix"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.12084975838661194,
          "label": "machine"
        },
        {
          "confidence": 0.120849609375,
          "label": "keypad"
        },
        {
          "confidence": 0.06274423003196716,
          "label": "tableware"
        },
        {
          "confidence": 0.06274423003196716,
          "label": "utensil"
        },
        {
          "confidence": 0.062744140625,
          "label": "plate"
        },
        {
          "confidence": 0.0298277847468853,
          "label": "structure"
        },
        {
          "confidence": 0.029824938625097275,
          "label": "sign"
        },
        {
          "confidence": 0.01882966235280037,
          "label": "material"
        },
        {
          "confidence": 0.018310546875,
          "label": "textile"
        },
        {
          "confidence": 0.014635302126407623,
          "label": "document"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "26"
      ],
      "text": "26"
    }
  }
}
Wikimedia source
Code 128 industrial barcode
barcode 2026 CC0 (locally generated)

Code 128 industrial barcode

Code 128 encodes ASCII letters and numbers - used on shipping labels, hospital wristbands, and warehouse pallets. Vision decodes all of these one-dimensional symbologies on-device.

$ auge --all code128-isbn.png
A OCR 1 line · 16 chars
AUGE-W1,1,0-PASS
# Classify 10 labels
art
13%
illustrations
13%
document
11%
chart
11%
diagram
11%
printed_page
7%
material
3%
tool
3%
matches
3%
structure
3%
Barcodes 1 found
Code128 AUGE-V1.1.0-PASS
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/code128-isbn.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "AUGE-V1.1.0-PASS",
          "symbology": "Code128"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.128173828125,
          "label": "art"
        },
        {
          "confidence": 0.128173828125,
          "label": "illustrations"
        },
        {
          "confidence": 0.10790302604436874,
          "label": "document"
        },
        {
          "confidence": 0.105224609375,
          "label": "chart"
        },
        {
          "confidence": 0.105224609375,
          "label": "diagram"
        },
        {
          "confidence": 0.07470703125,
          "label": "printed_page"
        },
        {
          "confidence": 0.030518805608153343,
          "label": "material"
        },
        {
          "confidence": 0.02807649038732052,
          "label": "tool"
        },
        {
          "confidence": 0.028076171875,
          "label": "matches"
        },
        {
          "confidence": 0.02569628693163395,
          "label": "structure"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "AUGE-W1,1,0-PASS"
      ],
      "text": "AUGE-W1,1,0-PASS"
    }
  }
}
Wikimedia source
QR code - WiFi network access
barcode 2026 CC0 (locally generated)

QR code - WiFi network access

Modern QR codes encode WiFi credentials in a single scan. auge --barcode decodes the WIFI: schema URI - on-device, never reaching out to the internet.

$ auge --all qr-wifi.png
A OCR no text
No text detected in this image.
# Classify 10 labels
art
7%
illustrations
7%
document
2%
structure
2%
sign
2%
chart
2%
diagram
2%
printed_page
2%
decoration
2%
frame
2%
Barcodes 1 found
QR WIFI:T:WPA;S:auge-network;P:on-device-only;H:false;;
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-wifi.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "WIFI:T:WPA;S:auge-network;P:on-device-only;H:false;;",
          "symbology": "QR"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.07327782362699509,
          "label": "art"
        },
        {
          "confidence": 0.0732421875,
          "label": "illustrations"
        },
        {
          "confidence": 0.021993102505803108,
          "label": "document"
        },
        {
          "confidence": 0.019250720739364624,
          "label": "structure"
        },
        {
          "confidence": 0.019044270738959312,
          "label": "sign"
        },
        {
          "confidence": 0.0185546875,
          "label": "chart"
        },
        {
          "confidence": 0.0185546875,
          "label": "diagram"
        },
        {
          "confidence": 0.018310546875,
          "label": "printed_page"
        },
        {
          "confidence": 0.01538088545203209,
          "label": "decoration"
        },
        {
          "confidence": 0.015380859375,
          "label": "frame"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
QR code - Bashō haiku
barcode 2026 CC0 (locally generated, haiku 1686 PD)

QR code - Bashō haiku

Matsuo Bashō's most famous haiku from 1686, encoded in a QR. auge decodes the multi-line text. The haiku itself is older than the United States.

$ auge --all qr-haiku.png
A OCR no text
No text detected in this image.
# Classify 10 labels
art
7%
illustrations
7%
document
4%
chart
4%
diagram
4%
printed_page
2%
material
2%
screenshot
2%
recreation
1%
puzzles
1%
Barcodes 1 found
QR An old silent pond A frog jumps into the pond - Splash! Silence again. - Matsuo Bashō, 1686
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-haiku.png",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": [
        {
          "payload": "An old silent pond\nA frog jumps into the pond -\nSplash! Silence again.\n- Matsuo Bashō, 1686",
          "symbology": "QR"
        }
      ]
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.0668945387005806,
          "label": "art"
        },
        {
          "confidence": 0.06689453125,
          "label": "illustrations"
        },
        {
          "confidence": 0.04280347377061844,
          "label": "document"
        },
        {
          "confidence": 0.0419921949505806,
          "label": "chart"
        },
        {
          "confidence": 0.0419921875,
          "label": "diagram"
        },
        {
          "confidence": 0.02099609375,
          "label": "printed_page"
        },
        {
          "confidence": 0.016358571127057076,
          "label": "material"
        },
        {
          "confidence": 0.01513671875,
          "label": "screenshot"
        },
        {
          "confidence": 0.010011732578277588,
          "label": "recreation"
        },
        {
          "confidence": 0.010009765625,
          "label": "puzzles"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Dürer's Rhinoceros (woodcut, 1515)
historical 1515 Public Domain

Dürer's Rhinoceros (woodcut, 1515)

Dürer's woodcut of an Indian rhinoceros - drawn from a written description, never seen in person. Above it: a Latin inscription. auge OCRs the inscription.

$ auge --all --enhance durer-rhinoceros.jpg
A OCR 4 lines · 371 chars
fang toot feyndt. Der welfant furcht es falt oßel/bant noo es Jit antumb/fo lauffe Ju das Ubier mit bem topff mijchen Oye fordern paynyond reyft ben selffanot onden ampauch auff
ñ erwirgt ]n/oes mag er fich nit erwer. Dann as Thier ift alfo gewapent/oas m ber belffanot nichts tan thün. Gie fagen auch basder 2bynocerus Schnell/ Sraydig ond Liffig fey.
ISIS
RHINOCERVS
# Classify 10 labels
art
93%
illustrations
93%
document
71%
printed_page
71%
animal
4%
mammal
4%
rhinoceros
4%
food
1%
vegetable
1%
pumpkin
1%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/durer-rhinoceros.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.92529296875,
          "label": "art"
        },
        {
          "confidence": 0.92529296875,
          "label": "illustrations"
        },
        {
          "confidence": 0.7060546875,
          "label": "document"
        },
        {
          "confidence": 0.7060546875,
          "label": "printed_page"
        },
        {
          "confidence": 0.03840738162398338,
          "label": "animal"
        },
        {
          "confidence": 0.03840675577521324,
          "label": "mammal"
        },
        {
          "confidence": 0.038330078125,
          "label": "rhinoceros"
        },
        {
          "confidence": 0.013672231696546078,
          "label": "food"
        },
        {
          "confidence": 0.013671927154064178,
          "label": "vegetable"
        },
        {
          "confidence": 0.013671875,
          "label": "pumpkin"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "fang toot feyndt. Der welfant furcht es falt oßel/bant noo es Jit antumb/fo lauffe Ju das Ubier mit bem topff mijchen Oye fordern paynyond reyft ben selffanot onden ampauch auff",
        "ñ erwirgt ]n/oes mag er fich nit erwer. Dann as Thier ift alfo gewapent/oas m ber belffanot nichts tan thün. Gie fagen auch basder 2bynocerus Schnell/ Sraydig ond Liffig fey.",
        "ISIS",
        "RHINOCERVS"
      ],
      "text": "fang toot feyndt. Der welfant furcht es falt oßel/bant noo es Jit antumb/fo lauffe Ju das Ubier mit bem topff mijchen Oye fordern paynyond reyft ben selffanot onden ampauch auff\nñ erwirgt ]n/oes mag er fich nit erwer. Dann as Thier ift alfo gewapent/oas m ber belffanot nichts tan thün. Gie fagen auch basder 2bynocerus Schnell/ Sraydig ond Liffig fey.\nISIS\nRHINOCERVS"
    }
  }
}
Wikimedia source
Gutenberg Bible page (c. 1455)
historical 1455 Public Domain

Gutenberg Bible page (c. 1455)

The first major book printed with movable type in Europe. Latin Vulgate in blackletter, two columns, illuminated. auge OCRs the Latin text with --langs la.

$ auge --all --langs la,en-US gutenberg-bible.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
document
43%
printed_page
43%
art
20%
illustrations
20%
book
14%
painting
9%
handwriting
4%
decoration
3%
frame
3%
structure
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/gutenberg-bible.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.4319077730178833,
          "label": "document"
        },
        {
          "confidence": 0.4306640625,
          "label": "printed_page"
        },
        {
          "confidence": 0.20235130190849304,
          "label": "art"
        },
        {
          "confidence": 0.199951171875,
          "label": "illustrations"
        },
        {
          "confidence": 0.142333984375,
          "label": "book"
        },
        {
          "confidence": 0.09375,
          "label": "painting"
        },
        {
          "confidence": 0.04052734375,
          "label": "handwriting"
        },
        {
          "confidence": 0.031755659729242325,
          "label": "decoration"
        },
        {
          "confidence": 0.03173828125,
          "label": "frame"
        },
        {
          "confidence": 0.03110364079475403,
          "label": "structure"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
United States Declaration of Independence, 1776
historical 1776 Public Domain

United States Declaration of Independence, 1776

The 1776 manuscript. Faded ink, dense colonial-era handwriting, irregular line spacing. auge runs every analysis at once.

$ auge --all --langs en-US --enhance declaration-of-independence.jpg
A OCR 79 lines · 3651 chars
IN CONGRESS, JULY 4, 1776.
The unanimous Selaration or be titom mito States of Hmerica.
bett in ticlours of human evendo it becoma necfoam for one pecoph to difolu the poltical bands which have connected them with another, and to -
yume among the flowers of the call, the jopardle and equat flation to which the Laus of Silene and of Natured lod ontitle them, a decent refect to the ofinions of mankind requies that thes
hould declare the caufes whick impu them tole fouralion.
We hail these truths tole felf-evident, that all more are create equal, that they are endowed by the brea
with calain unalunable Righls, thatamong these we Life, Lito and the pursuit of Happinep.—
That lo feeure these lights, bovenments are instituted among Men, derring the just
Tal wheneve any form of boren ment becomes destructeve of these ondo, it is the Right ofthe People lo alla orto abolish it, and to instilate neu
yorenment, laying it foundation on such fencifies and organizing its fones in such form, aste them shall sum most likely to effed the dely and Slappings. Prudence, ended
wold.
and preping unportana, unto fuspended inther operation tell his dont shoud be obtained; and when so fuspended, hehas allely negleted to allnd lolhem
fags other Laus for the accommodation of large distrids of people, untep those keople would icingush the right of Refusentation in the Legistase, a right ineffimable lolhem and formidable
to tyrants only.
compliance with his merfues.
and payment ofter salames.
us, in times of fease, Standing Ahmus without the consent of our legislatimes.
will others tapelpot us to a justicion forign lo our constillion, and unacknowleged by our laws; giving his Alient lotte lets of feelended Legitation:
He has comerne
İs quatring large toding
so as to render it atone an coample and fit infument for inhoduang thesame absolule rule into these Colonies: -
Hor taking away one harles, abolishing one most valuable laus, and
Jar spending our oun Legislates, and delaring themslus inveiled wite foun lo legislate for us in all cans vhulioure.-
He he abdicatet lyonanment hoe, by delaing as out ofles Poletion and rooging Ma agains us. — O He has flinted our sas, caraged one boads, buent our towns and diskoyed the din
фон /copt.
He is at this time kansferting large lomics of fougn Mercenares to complet the works of death, deselation and tranny, abeady begun win cucumstances of ruelly spapa
He has constained our fellow bilizens taken baflie onthe high Seas lobear lirns again
Herbounky, lobecame the cacutions offer finds and Bechun, aslofall thematoes by this Hands. -
He has recited domeste infuredions amongst us, and has endeavowed lo bung orell
Ike have eminded them of the circumstancs four omigration and fellement here. We have appeded tother native justic and magnanimily, and we have conjined then
Shey too have been deaf to the voice of justice and of
consanguinity.
We must, therefore, acquiesce in the necefit, which denounes our Separation, and hold them, as we hold the ust ofmankind, Cremis in He, in Peace Frunas.
Mates may of right do.
and our faced Honor
Button Quennsts
dyman Heal,
GeoMacion.
Про Коорет
SauplAewes,
Soh.. Pinn
Edward Rathedge /
The legues, gunt
Thomas did fur
Athur Madleions
Janual Plane,
10m Paca
Hot Fon
Ковтоті
B'enjaminiusn
Jun Morton
Lvia.
res
George Vagthe
Richard Henry 2e.
Haroone
Carter Braxton ..
James Wilson
Tech Porklon
C0: 7108
Coolien
HoI. Frate
Tra! Clophenson
John lent
Abra Clark
Tosiah Bartlito
B? Tipple
Jani Adame
John A dama
Rot trav aunte
Ilbridy
Jerry
Step Hopkins
William Clamp s
Proger Sherman
dam" Huntinglan
My litiam,
Oliver Not
Matthew Thowton
# Classify 8 labels
document
83%
printed_page
81%
handwriting
51%
art
3%
illustrations
3%
newspaper
2%
book
1%
material
1%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/declaration-of-independence.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8279598355293274,
          "label": "document"
        },
        {
          "confidence": 0.81005859375,
          "label": "printed_page"
        },
        {
          "confidence": 0.5126953125,
          "label": "handwriting"
        },
        {
          "confidence": 0.026855487376451492,
          "label": "art"
        },
        {
          "confidence": 0.02685546875,
          "label": "illustrations"
        },
        {
          "confidence": 0.017822265625,
          "label": "newspaper"
        },
        {
          "confidence": 0.01220703125,
          "label": "book"
        },
        {
          "confidence": 0.010257808491587639,
          "label": "material"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "IN CONGRESS, JULY 4, 1776.",
        "The unanimous Selaration or be titom mito States of Hmerica.",
        "bett in ticlours of human evendo it becoma necfoam for one pecoph to difolu the poltical bands which have connected them with another, and to -",
        "yume among the flowers of the call, the jopardle and equat flation to which the Laus of Silene and of Natured lod ontitle them, a decent refect to the ofinions of mankind requies that thes",
        "hould declare the caufes whick impu them tole fouralion.",
        "We hail these truths tole felf-evident, that all more are create equal, that they are endowed by the brea",
        "with calain unalunable Righls, thatamong these we Life, Lito and the pursuit of Happinep.—",
        "That lo feeure these lights, bovenments are instituted among Men, derring the just",
        "Tal wheneve any form of boren ment becomes destructeve of these ondo, it is the Right ofthe People lo alla orto abolish it, and to instilate neu",
        "yorenment, laying it foundation on such fencifies and organizing its fones in such form, aste them shall sum most likely to effed the dely and Slappings. Prudence, ended",
        "wold.",
        "and preping unportana, unto fuspended inther operation tell his dont shoud be obtained; and when so fuspended, hehas allely negleted to allnd lolhem",
        "fags other Laus for the accommodation of large distrids of people, untep those keople would icingush the right of Refusentation in the Legistase, a right ineffimable lolhem and formidable",
        "to tyrants only.",
        "compliance with his merfues.",
        "and payment ofter salames.",
        "us, in times of fease, Standing Ahmus without the consent of our legislatimes.",
        "will others tapelpot us to a justicion forign lo our constillion, and unacknowleged by our laws; giving his Alient lotte lets of feelended Legitation:",
        "He has comerne",
        "İs quatring large toding",
        "so as to render it atone an coample and fit infument for inhoduang thesame absolule rule into these Colonies: -",
        "Hor taking away one harles, abolishing one most valuable laus, and",
        "Jar spending our oun Legislates, and delaring themslus inveiled wite foun lo legislate for us in all cans vhulioure.-",
        "He he abdicatet lyonanment hoe, by delaing as out ofles Poletion and rooging Ma agains us. — O He has flinted our sas, caraged one boads, buent our towns and diskoyed the din",
        "фон /copt.",
        "He is at this time kansferting large lomics of fougn Mercenares to complet the works of death, deselation and tranny, abeady begun win cucumstances of ruelly spapa",
        "He has constained our fellow bilizens taken baflie onthe high Seas lobear lirns again",
        "Herbounky, lobecame the cacutions offer finds and Bechun, aslofall thematoes by this Hands. -",
        "He has recited domeste infuredions amongst us, and has endeavowed lo bung orell",
        "Ike have eminded them of the circumstancs four omigration and fellement here. We have appeded tother native justic and magnanimily, and we have conjined then",
        "Shey too have been deaf to the voice of justice and of",
        "consanguinity.",
        "We must, therefore, acquiesce in the necefit, which denounes our Separation, and hold them, as we hold the ust ofmankind, Cremis in He, in Peace Frunas.",
        "Mates may of right do.",
        "and our faced Honor",
        "Button Quennsts",
        "dyman Heal,",
        "GeoMacion.",
        "Про Коорет",
        "SauplAewes,",
        "Soh.. Pinn",
        "Edward Rathedge /",
        "The legues, gunt",
        "Thomas did fur",
        "Athur Madleions",
        "Janual Plane,",
        "10m Paca",
        "Hot Fon",
        "Ковтоті",
        "B'enjaminiusn",
        "Jun Morton",
        "Lvia.",
        "res",
        "George Vagthe",
        "Richard Henry 2e.",
        "Haroone",
        "Carter Braxton ..",
        "James Wilson",
        "Tech Porklon",
        "C0: 7108",
        "Coolien",
        "HoI. Frate",
        "Tra! Clophenson",
        "John lent",
        "Abra Clark",
        "Tosiah Bartlito",
        "B? Tipple",
        "Jani Adame",
        "John A dama",
        "Rot trav aunte",
        "Ilbridy",
        "Jerry",
        "Step Hopkins",
        "William Clamp s",
        "Proger Sherman",
        "dam\" Huntinglan",
        "My litiam,",
        "Oliver Not",
        "Matthew Thowton"
      ],
      "text": "IN CONGRESS, JULY 4, 1776.\nThe unanimous Selaration or be titom mito States of Hmerica.\nbett in ticlours of human evendo it becoma necfoam for one pecoph to difolu the poltical bands which have connected them with another, and to -\nyume among the flowers of the call, the jopardle and equat flation to which the Laus of Silene and of Natured lod ontitle them, a decent refect to the ofinions of mankind requies that thes\nhould declare the caufes whick impu them tole fouralion.\nWe hail these truths tole felf-evident, that all more are create equal, that they are endowed by the brea\nwith calain unalunable Righls, thatamong these we Life, Lito and the pursuit of Happinep.—\nThat lo feeure these lights, bovenments are instituted among Men, derring the just\nTal wheneve any form of boren ment becomes destructeve of these ondo, it is the Right ofthe People lo alla orto abolish it, and to instilate neu\nyorenment, laying it foundation on such fencifies and organizing its fones in such form, aste them shall sum most likely to effed the dely and Slappings. Prudence, ended\nwold.\nand preping unportana, unto fuspended inther operation tell his dont shoud be obtained; and when so fuspended, hehas allely negleted to allnd lolhem\nfags other Laus for the accommodation of large distrids of people, untep those keople would icingush the right of Refusentation in the Legistase, a right ineffimable lolhem and formidable\nto tyrants only.\ncompliance with his merfues.\nand payment ofter salames.\nus, in times of fease, Standing Ahmus without the consent of our legislatimes.\nwill others tapelpot us to a justicion forign lo our constillion, and unacknowleged by our laws; giving his Alient lotte lets of feelended Legitation:\nHe has comerne\nİs quatring large toding\nso as to render it atone an coample and fit infument for inhoduang thesame absolule rule into these Colonies: -\nHor taking away one harles, abolishing one most valuable laus, and\nJar spending our oun Legislates, and delaring themslus inveiled wite foun lo legislate for us in all cans vhulioure.-\nHe he abdicatet lyonanment hoe, by delaing as out ofles Poletion and rooging Ma agains us. — O He has flinted our sas, caraged one boads, buent our towns and diskoyed the din\nфон /copt.\nHe is at this time kansferting large lomics of fougn Mercenares to complet the works of death, deselation and tranny, abeady begun win cucumstances of ruelly spapa\nHe has constained our fellow bilizens taken baflie onthe high Seas lobear lirns again\nHerbounky, lobecame the cacutions offer finds and Bechun, aslofall thematoes by this Hands. -\nHe has recited domeste infuredions amongst us, and has endeavowed lo bung orell\nIke have eminded them of the circumstancs four omigration and fellement here. We have appeded tother native justic and magnanimily, and we have conjined then\nShey too have been deaf to the voice of justice and of\nconsanguinity.\nWe must, therefore, acquiesce in the necefit, which denounes our Separation, and hold them, as we hold the ust ofmankind, Cremis in He, in Peace Frunas.\nMates may of right do.\nand our faced Honor\nButton Quennsts\ndyman Heal,\nGeoMacion.\nПро Коорет\nSauplAewes,\nSoh.. Pinn\nEdward Rathedge /\nThe legues, gunt\nThomas did fur\nAthur Madleions\nJanual Plane,\n10m Paca\nHot Fon\nКовтоті\nB'enjaminiusn\nJun Morton\nLvia.\nres\nGeorge Vagthe\nRichard Henry 2e.\nHaroone\nCarter Braxton ..\nJames Wilson\nTech Porklon\nC0: 7108\nCoolien\nHoI. Frate\nTra! Clophenson\nJohn lent\nAbra Clark\nTosiah Bartlito\nB? Tipple\nJani Adame\nJohn A dama\nRot trav aunte\nIlbridy\nJerry\nStep Hopkins\nWilliam Clamp s\nProger Sherman\ndam\" Huntinglan\nMy litiam,\nOliver Not\nMatthew Thowton"
    }
  }
}
Wikimedia source
James Lyne's New York Survey (Bradford map), c. 1730
historical 1730 Public Domain

James Lyne's New York Survey (Bradford map), c. 1730

An 18th-century map of New York City. Tiny labels, decorative typefaces. Real adversarial input - --enhance upscales tiny lines.

$ auge --all --enhance ny-bradford-map.jpg
A OCR 4 lines · 35 chars
PALISADES
ED HOLLAVD F/
MAYORY
11 A
# Classify 6 labels
document
88%
map
88%
art
31%
illustrations
31%
printed_page
5%
handwriting
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/ny-bradford-map.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8750012516975403,
          "label": "document"
        },
        {
          "confidence": 0.875,
          "label": "map"
        },
        {
          "confidence": 0.30810546875,
          "label": "art"
        },
        {
          "confidence": 0.30810546875,
          "label": "illustrations"
        },
        {
          "confidence": 0.051025390625,
          "label": "printed_page"
        },
        {
          "confidence": 0.03271484375,
          "label": "handwriting"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "PALISADES",
        "ED HOLLAVD F/",
        "MAYORY",
        "11 A"
      ],
      "text": "PALISADES\nED HOLLAVD F/\nMAYORY\n11 A"
    }
  }
}
Wikimedia source
The Rosetta Stone (196 BC)
multilingual -196 Public Domain

The Rosetta Stone (196 BC)

The same decree in three scripts - hieroglyphic, demotic Egyptian, ancient Greek. The original multilingual document. auge OCRs the visible Greek section with --langs el-GR.

$ auge --all --langs el-GR,en-US --enhance rosetta-stone.jpg
A OCR 2 lines · 14 chars
amples
201e-1a
# Classify 10 labels
container
10%
luggage
10%
purse
10%
material
9%
textile
9%
clothing
4%
hat
3%
headgear
3%
fedora
3%
jeans
3%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/rosetta-stone.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.09890684485435486,
          "label": "container"
        },
        {
          "confidence": 0.09890399128198624,
          "label": "luggage"
        },
        {
          "confidence": 0.098876953125,
          "label": "purse"
        },
        {
          "confidence": 0.09033213555812836,
          "label": "material"
        },
        {
          "confidence": 0.09033203125,
          "label": "textile"
        },
        {
          "confidence": 0.03598817065358162,
          "label": "clothing"
        },
        {
          "confidence": 0.03205922991037369,
          "label": "hat"
        },
        {
          "confidence": 0.03205922991037369,
          "label": "headgear"
        },
        {
          "confidence": 0.031494140625,
          "label": "fedora"
        },
        {
          "confidence": 0.02734375,
          "label": "jeans"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "amples",
        "201e-1a"
      ],
      "text": "amples\n201e-1a"
    }
  }
}
Wikimedia source
Magna Carta - 1215, British Library copy
historical 1215 Public Domain

Magna Carta - 1215, British Library copy

The 1215 charter that limited the English king's power. Medieval Latin script in dense lines - extreme OCR difficulty. Apple's model attempts it anyway.

$ auge --all --enhance magna-carta.jpg
A OCR 68 lines · 1758 chars
Kintadctcen falafe
Wideroli ns
ST.H.
adastant t
cefallf
S.E.M
-0-web.Mur afec coe)
S. C.
come paliy
mt-afaf mocn)er oml nos ti
-5 fac
1 s-foa Bain Ge
Rotatia
C.A.
Kove faned Ceef
N10-0-
e Cm:
•Grase flore Srot atlem Lane Lol
a gdr deetn po ite seden
•d sid poll eee
облетів.
DateTanmhaol-cics
Mana A
eus Bielia
falice
Pitec fofm den Crucfo Cwe 251803 1 r
paT.....lsihalumfsffet danes Glodeat
he rone aha ahth.nevn
and. An farso CufL
laten
fe rontlerd
Are ale n fee pofta abro poset a andasort toe ate fomdepore- amfen iSL.
As li Canos....
onen t.lA.
sNL hm Hal.t.d solo boloho peformAnt t, silaeflenedbué a horenity
mala mesun na fosn
wssldtbslL.ieminttietahotkallte.mesafamd.10-lemmnt._Caerbolt
a tone Be cen sandae atea d loom gonferp blene festpm ahe uf fo fuf honw ae waf-indeasalen stear ale mor
sedhple
•Cà prese
mi nt En
sefe a pd «lore - l.
h floc har ahahn foa afle
Votesa
oc rosestenem item mis
hetafa-af fordilebzngunpos0-mn aof cfolsjrnder
nce neof comet2 aloe metem chel
ceafendem acat s fe welnth
t folloses
e Fenta ioteefines o
aF6 Cim a Pigmn 1.0he mea (00
SiraFrp coadSmeed ep
Soroa pati fa alan Ammewer ca Bim Dalsus tomagang Bomfe me, Frondana
ah poar Cif p Antren llies it-um Slmm-1
tr timee suite
-sinapf-forfet-ather wee hpams gump Barelh Arngods- gRade wufeeef Grocd medee dle nanddee hephed Chsle Spear
yi Borenh doaveshom lacen giraio fue d
mAke niMs
e conch tr ane ihe fame
.Streml: cmene oxquadt f ferpn
.fototaalin tilih.-Sond ace con
S.c Coene eanw/mehn alamof aleeed1.hrmmcen
Ate ninea Fanust
ar fese nrefone ned-sofend a Datk Ime femmems Sanome nefaadd phcevefendehm plenes
(cEro stas te anr Dandull supkfiche
e ana foa fee nalo mgene deres.1
malae Mes
amhmlachd cosanda nfor hani sando dada hea D CoSa trahocle
# Classify 7 labels
document
31%
handwriting
31%
material
12%
textile
12%
printed_page
8%
newspaper
8%
book
4%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/magna-carta.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.3125331401824951,
          "label": "document"
        },
        {
          "confidence": 0.311767578125,
          "label": "handwriting"
        },
        {
          "confidence": 0.12330673635005951,
          "label": "material"
        },
        {
          "confidence": 0.123291015625,
          "label": "textile"
        },
        {
          "confidence": 0.083984375,
          "label": "printed_page"
        },
        {
          "confidence": 0.08056640625,
          "label": "newspaper"
        },
        {
          "confidence": 0.036865234375,
          "label": "book"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "Kintadctcen falafe",
        "Wideroli ns",
        "ST.H.",
        "adastant t",
        "cefallf",
        "S.E.M",
        "-0-web.Mur afec coe)",
        "S. C.",
        "come paliy",
        "mt-afaf mocn)er oml nos ti",
        "-5 fac",
        "1 s-foa Bain Ge",
        "Rotatia",
        "C.A.",
        "Kove faned Ceef",
        "N10-0-",
        "e Cm:",
        "•Grase flore Srot atlem Lane Lol",
        "a gdr deetn po ite seden",
        "•d sid poll eee",
        "облетів.",
        "DateTanmhaol-cics",
        "Mana A",
        "eus Bielia",
        "falice",
        "Pitec fofm den Crucfo Cwe 251803 1 r",
        "paT.....lsihalumfsffet danes Glodeat",
        "he rone aha ahth.nevn",
        "and. An farso CufL",
        "laten",
        "fe rontlerd",
        "Are ale n fee pofta abro poset a andasort toe ate fomdepore- amfen iSL.",
        "As li Canos....",
        "onen t.lA.",
        "sNL hm Hal.t.d solo boloho peformAnt t, silaeflenedbué a horenity",
        "mala mesun na fosn",
        "wssldtbslL.ieminttietahotkallte.mesafamd.10-lemmnt._Caerbolt",
        "a tone Be cen sandae atea d loom gonferp blene festpm ahe uf fo fuf honw ae waf-indeasalen stear ale mor",
        "sedhple",
        "•Cà prese",
        "mi nt En",
        "sefe a pd «lore - l.",
        "h floc har ahahn foa afle",
        "Votesa",
        "oc rosestenem item mis",
        "hetafa-af fordilebzngunpos0-mn aof cfolsjrnder",
        "nce neof comet2 aloe metem chel",
        "ceafendem acat s fe welnth",
        "t folloses",
        "e Fenta ioteefines o",
        "aF6 Cim a Pigmn 1.0he mea (00",
        "SiraFrp coadSmeed ep",
        "Soroa pati fa alan Ammewer ca Bim Dalsus tomagang Bomfe me, Frondana",
        "ah poar Cif p Antren llies it-um Slmm-1",
        "tr timee suite",
        "-sinapf-forfet-ather wee hpams gump Barelh Arngods- gRade wufeeef Grocd medee dle nanddee hephed Chsle Spear",
        "yi Borenh doaveshom lacen giraio fue d",
        "mAke niMs",
        "e conch tr ane ihe fame",
        ".Streml: cmene oxquadt f ferpn",
        ".fototaalin tilih.-Sond ace con",
        "S.c Coene eanw/mehn alamof aleeed1.hrmmcen",
        "Ate ninea Fanust",
        "ar fese nrefone ned-sofend a Datk Ime femmems Sanome nefaadd phcevefendehm plenes",
        "(cEro stas te anr Dandull supkfiche",
        "e ana foa fee nalo mgene deres.1",
        "malae Mes",
        "amhmlachd cosanda nfor hani sando dada hea D CoSa trahocle"
      ],
      "text": "Kintadctcen falafe\nWideroli ns\nST.H.\nadastant t\ncefallf\nS.E.M\n-0-web.Mur afec coe)\nS. C.\ncome paliy\nmt-afaf mocn)er oml nos ti\n-5 fac\n1 s-foa Bain Ge\nRotatia\nC.A.\nKove faned Ceef\nN10-0-\ne Cm:\n•Grase flore Srot atlem Lane Lol\na gdr deetn po ite seden\n•d sid poll eee\nоблетів.\nDateTanmhaol-cics\nMana A\neus Bielia\nfalice\nPitec fofm den Crucfo Cwe 251803 1 r\npaT.....lsihalumfsffet danes Glodeat\nhe rone aha ahth.nevn\nand. An farso CufL\nlaten\nfe rontlerd\nAre ale n fee pofta abro poset a andasort toe ate fomdepore- amfen iSL.\nAs li Canos....\nonen t.lA.\nsNL hm Hal.t.d solo boloho peformAnt t, silaeflenedbué a horenity\nmala mesun na fosn\nwssldtbslL.ieminttietahotkallte.mesafamd.10-lemmnt._Caerbolt\na tone Be cen sandae atea d loom gonferp blene festpm ahe uf fo fuf honw ae waf-indeasalen stear ale mor\nsedhple\n•Cà prese\nmi nt En\nsefe a pd «lore - l.\nh floc har ahahn foa afle\nVotesa\noc rosestenem item mis\nhetafa-af fordilebzngunpos0-mn aof cfolsjrnder\nnce neof comet2 aloe metem chel\nceafendem acat s fe welnth\nt folloses\ne Fenta ioteefines o\naF6 Cim a Pigmn 1.0he mea (00\nSiraFrp coadSmeed ep\nSoroa pati fa alan Ammewer ca Bim Dalsus tomagang Bomfe me, Frondana\nah poar Cif p Antren llies it-um Slmm-1\ntr timee suite\n-sinapf-forfet-ather wee hpams gump Barelh Arngods- gRade wufeeef Grocd medee dle nanddee hephed Chsle Spear\nyi Borenh doaveshom lacen giraio fue d\nmAke niMs\ne conch tr ane ihe fame\n.Streml: cmene oxquadt f ferpn\n.fototaalin tilih.-Sond ace con\nS.c Coene eanw/mehn alamof aleeed1.hrmmcen\nAte ninea Fanust\nar fese nrefone ned-sofend a Datk Ime femmems Sanome nefaadd phcevefendehm plenes\n(cEro stas te anr Dandull supkfiche\ne ana foa fee nalo mgene deres.1\nmalae Mes\namhmlachd cosanda nfor hani sando dada hea D CoSa trahocle"
    }
  }
}
Wikimedia source
The Last Supper - Leonardo da Vinci, 1498
faces 1498 Public Domain

The Last Supper - Leonardo da Vinci, 1498

Leonardo's mural in the refectory of Santa Maria delle Grazie, Milan. Thirteen named figures arranged with mathematical precision. auge counts the apostles.

$ auge --all last-supper.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
people
61%
adult
61%
art
55%
painting
55%
structure
21%
arch
20%
material
9%
textile
9%
child
7%
crowd
7%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/last-supper.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6055296659469604,
          "label": "people"
        },
        {
          "confidence": 0.60546875,
          "label": "adult"
        },
        {
          "confidence": 0.5464094281196594,
          "label": "art"
        },
        {
          "confidence": 0.54638671875,
          "label": "painting"
        },
        {
          "confidence": 0.21249714493751526,
          "label": "structure"
        },
        {
          "confidence": 0.2001953125,
          "label": "arch"
        },
        {
          "confidence": 0.08988814800977707,
          "label": "material"
        },
        {
          "confidence": 0.08984375,
          "label": "textile"
        },
        {
          "confidence": 0.074951171875,
          "label": "child"
        },
        {
          "confidence": 0.07373046875,
          "label": "crowd"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The School of Athens - Raphael, 1511
faces 1511 Public Domain

The School of Athens - Raphael, 1511

Raphael's 1511 fresco in the Vatican: Plato, Aristotle, Socrates, Pythagoras, Euclid, Diogenes - twenty-some philosophers in one room. auge counts the faces and classifies the architecture.

$ auge --all raphael-athens.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
art
82%
painting
82%
structure
78%
arch
78%
people
39%
crowd
36%
adult
28%
statue
22%
interior_room
14%
museum
13%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/raphael-athens.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8208773136138916,
          "label": "art"
        },
        {
          "confidence": 0.81787109375,
          "label": "painting"
        },
        {
          "confidence": 0.7828600406646729,
          "label": "structure"
        },
        {
          "confidence": 0.7822265625,
          "label": "arch"
        },
        {
          "confidence": 0.39246267080307007,
          "label": "people"
        },
        {
          "confidence": 0.364990234375,
          "label": "crowd"
        },
        {
          "confidence": 0.279296875,
          "label": "adult"
        },
        {
          "confidence": 0.224609375,
          "label": "statue"
        },
        {
          "confidence": 0.13782037794589996,
          "label": "interior_room"
        },
        {
          "confidence": 0.130126953125,
          "label": "museum"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Night Watch - Rembrandt, 1642
faces 1642 Public Domain

The Night Watch - Rembrandt, 1642

Rembrandt's monumental civic guard portrait at the Rijksmuseum. Eighteen named figures plus several extras. A genuine stress test for face detection on chiaroscuro paintings.

$ auge --all rembrandt-nightwatch.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
people
39%
adult
36%
crowd
29%
art
26%
painting
26%
toy
20%
figurine
20%
tool
12%
celebration
10%
costume
10%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/rembrandt-nightwatch.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.39403674006462097,
          "label": "people"
        },
        {
          "confidence": 0.360595703125,
          "label": "adult"
        },
        {
          "confidence": 0.2919921875,
          "label": "crowd"
        },
        {
          "confidence": 0.26047685742378235,
          "label": "art"
        },
        {
          "confidence": 0.259033203125,
          "label": "painting"
        },
        {
          "confidence": 0.20241057872772217,
          "label": "toy"
        },
        {
          "confidence": 0.202392578125,
          "label": "figurine"
        },
        {
          "confidence": 0.1242794618010521,
          "label": "tool"
        },
        {
          "confidence": 0.1044926717877388,
          "label": "celebration"
        },
        {
          "confidence": 0.1044921875,
          "label": "costume"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Wanderer above the Sea of Fog - Caspar David Friedrich, 1818
classify 1818 Public Domain

Wanderer above the Sea of Fog - Caspar David Friedrich, 1818

The defining image of German Romanticism. A solitary figure above mountain mist - auge tries face detection (back of head, no face visible) and classify on a Romantic landscape.

$ auge --all friedrich-wanderer.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
outdoor
61%
sky
61%
cloudy
61%
rocks
50%
structure
50%
people
41%
adult
41%
land
19%
hill
19%
liquid
13%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/friedrich-wanderer.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.6129744648933411,
          "label": "outdoor"
        },
        {
          "confidence": 0.6118281483650208,
          "label": "sky"
        },
        {
          "confidence": 0.61181640625,
          "label": "cloudy"
        },
        {
          "confidence": 0.5009765625,
          "label": "rocks"
        },
        {
          "confidence": 0.5009765625,
          "label": "structure"
        },
        {
          "confidence": 0.4108887016773224,
          "label": "people"
        },
        {
          "confidence": 0.410888671875,
          "label": "adult"
        },
        {
          "confidence": 0.1870730072259903,
          "label": "land"
        },
        {
          "confidence": 0.186767578125,
          "label": "hill"
        },
        {
          "confidence": 0.12943993508815765,
          "label": "liquid"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
The Solvay Conference, Brussels 1927
face 1
face 2
face 3
face 4
face 5
face 6
faces 1927 Public Domain (anonymous, 70+ years)

The Solvay Conference, Brussels 1927

Twenty-nine of the 20th century's greatest physicists in one frame: Einstein, Curie, Bohr, Heisenberg, Schrödinger, Dirac, Pauli, Lorentz. auge --faces detects and locates them.

$ auge --all solvay-1927.jpg
A OCR 1 line · 3 chars
ROD
# Classify 10 labels
clothing
70%
people
69%
suit
65%
adult
65%
military_uniform
52%
crowd
49%
structure
34%
conveyance
34%
portal
34%
window
34%
Barcodes none
No barcodes or QR codes detected.
Faces 6 detected
6 faces located. Bounding boxes overlaid on the image above.
face 1: x=0.855 y=0.453 w=0.035 h=0.049
face 2: x=0.137 y=0.453 w=0.032 h=0.045
face 3: x=0.347 y=0.467 w=0.035 h=0.048
face 4: x=0.815 y=0.415 w=0.035 h=0.049
face 5: x=0.285 y=0.423 w=0.032 h=0.044
face 6: x=0.756 y=0.461 w=0.033 h=0.046
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/solvay-1927.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7000404596328735,
          "label": "clothing"
        },
        {
          "confidence": 0.6874275803565979,
          "label": "people"
        },
        {
          "confidence": 0.65087890625,
          "label": "suit"
        },
        {
          "confidence": 0.6484375,
          "label": "adult"
        },
        {
          "confidence": 0.521484375,
          "label": "military_uniform"
        },
        {
          "confidence": 0.487548828125,
          "label": "crowd"
        },
        {
          "confidence": 0.3414841294288635,
          "label": "structure"
        },
        {
          "confidence": 0.341309517621994,
          "label": "conveyance"
        },
        {
          "confidence": 0.3413088321685791,
          "label": "portal"
        },
        {
          "confidence": 0.34130859375,
          "label": "window"
        }
      ]
    },
    "faces": {
      "count": 6,
      "faces": [
        {
          "height": 0.04886019974946976,
          "width": 0.03534730151295662,
          "x": 0.8549006581306458,
          "y": 0.4525025188922882
        },
        {
          "height": 0.04458567500114441,
          "width": 0.03225494921207428,
          "x": 0.13719181716442108,
          "y": 0.4527735412120819
        },
        {
          "height": 0.04770072549581528,
          "width": 0.034508492797613144,
          "x": 0.34748557209968567,
          "y": 0.4667179584503174
        },
        {
          "height": 0.048947274684906006,
          "width": 0.03541029244661331,
          "x": 0.8145434260368347,
          "y": 0.4153168201446533
        },
        {
          "height": 0.04424234852194786,
          "width": 0.03200657293200493,
          "x": 0.2846284508705139,
          "y": 0.42316147685050964
        },
        {
          "height": 0.04566596820950508,
          "width": 0.033036474138498306,
          "x": 0.755698561668396,
          "y": 0.4610947370529175
        }
      ]
    },
    "ocr": {
      "lines": [
        "ROD"
      ],
      "text": "ROD"
    }
  }
}
Wikimedia source
Chinese road sign with QR code (Hainan, outdoor)
barcode 2025 Public Domain / CC0

Chinese road sign with QR code (Hainan, outdoor)

A real outdoor QR code in the wild - on a Chinese highway sign. auge decodes the URL plus extracts the Mandarin text alongside it. One image, every analysis.

$ auge --all --langs zh-Hans,en-US qr-outdoor-china.jpg
A OCR 4 lines · 51 chars
扫码
预约通行
Scan OR codes
Reserv ation pass
# Classify 10 labels
outdoor
87%
sky
87%
blue_sky
79%
structure
75%
cloudy
73%
sign
65%
street_sign
64%
fence
63%
land
43%
grass
40%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/qr-outdoor-china.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.8734795451164246,
          "label": "outdoor"
        },
        {
          "confidence": 0.8677766919136047,
          "label": "sky"
        },
        {
          "confidence": 0.7919921875,
          "label": "blue_sky"
        },
        {
          "confidence": 0.7460048198699951,
          "label": "structure"
        },
        {
          "confidence": 0.7314453125,
          "label": "cloudy"
        },
        {
          "confidence": 0.6533144116401672,
          "label": "sign"
        },
        {
          "confidence": 0.64404296875,
          "label": "street_sign"
        },
        {
          "confidence": 0.62841796875,
          "label": "fence"
        },
        {
          "confidence": 0.43182459473609924,
          "label": "land"
        },
        {
          "confidence": 0.403076171875,
          "label": "grass"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [
        "扫码",
        "预约通行",
        "Scan OR codes",
        "Reserv ation pass"
      ],
      "text": "扫码\n预约通行\nScan OR codes\nReserv ation pass"
    }
  }
}
Wikimedia source
Dignity and Impudence - Edwin Landseer, 1839
classify 1839 Public Domain

Dignity and Impudence - Edwin Landseer, 1839

A sad-eyed bloodhound and a cheeky terrier. Landseer was Queen Victoria's favourite painter. auge classifies the dogs (very accurately) - and may try face-detection on dog faces.

$ auge --all landseer-dogs.jpg
A OCR no text
No text detected in this image.
# Classify 10 labels
structure
75%
wood_processed
75%
animal
15%
canine
15%
dog
15%
mammal
15%
ridgeback
14%
arch
13%
vizsla
9%
dachshund
9%
Barcodes none
No barcodes or QR codes detected.
Faces 0 detected
No faces detected.
{} Raw JSON on-device output
{
  "file": "/Users/arthurficial/dev/auge-web/corpus/files/landseer-dogs.jpg",
  "metadata": {
    "on_device": true,
    "version": "1.1.0"
  },
  "mode": "all",
  "results": {
    "barcodes": {
      "barcodes": []
    },
    "classify": {
      "classifications": [
        {
          "confidence": 0.7452318668365479,
          "label": "structure"
        },
        {
          "confidence": 0.7451171875,
          "label": "wood_processed"
        },
        {
          "confidence": 0.1506083756685257,
          "label": "animal"
        },
        {
          "confidence": 0.1506083756685257,
          "label": "canine"
        },
        {
          "confidence": 0.1506083756685257,
          "label": "dog"
        },
        {
          "confidence": 0.1506083756685257,
          "label": "mammal"
        },
        {
          "confidence": 0.137451171875,
          "label": "ridgeback"
        },
        {
          "confidence": 0.128662109375,
          "label": "arch"
        },
        {
          "confidence": 0.091552734375,
          "label": "vizsla"
        },
        {
          "confidence": 0.091064453125,
          "label": "dachshund"
        }
      ]
    },
    "faces": {
      "count": 0,
      "faces": []
    },
    "ocr": {
      "lines": [],
      "text": ""
    }
  }
}
Wikimedia source
Install

Get auge.

Two install paths. Pick one. Both run the same binary.

Homebrew (recommended)
brew tap Arthur-Ficial/tap
brew install Arthur-Ficial/tap/auge
Build from source
git clone https://github.com/Arthur-Ficial/auge
cd auge && make install

No Xcode required. Builds with Command Line Tools. macOS 10.15+.

The apfel tree

Apple's on-device intelligence, exposed as UNIX tools.

Each tool wraps one Apple framework. Together they are the apfel tree - small, focused CLIs for the intelligence already shipping on your Mac.