Explorer
Syncing—

Optical Character Recognition in Images - Google Cloud Vision compatible APIService

optical-character-recognition

Google Cloud Vision-compatible OCR (text detection and recognition). The expected endpoints are POST /v1/images:annotate (features TEXT_DETECTION and DOCUMENT_TEXT_DETECTION, REST only) and GET /health; suppliers MAY expose more of the Cloud Vision surface or extensions but consumers MUST NOT depend on it. This service is model-agnostic: suppliers choose their own OCR model and weights, so recognized text, language coverage, and polygons vary and are not guaranteed by the protocol. Every response, including errors, is a single JSON object.

Active Suppliers
5
of 5 ever configured
Service Info
Service ID
optical-character-recognition
Compute Units / Relay
50,000
Owner
pokt1l…2h0l
Active Applications
1
Relay Mining EMA
1 relays
Service Metadata Cardsource: on-chain metadata (LCD)
{9 items
"schema":"pocket-service-card/v1"
"description":"Google Cloud Vision-compatible OCR (text detection and recognition). The expected endpoints are POST /v1/images:annotate (features TEXT_DETECTION and DOCUMENT_T..."
"rpc_types":[1 item
0:{4 items
"type":"REST"
"intent":"expected"
"backend_hint":"Cloud Vision-compatible images:annotate server on :8080 (image-inference-server, ocr endpoint enabled); mount at /"
"notes":"Only POST /v1/images:annotate and GET /health are expected. Request body is Cloud Vision JSON: requests[{image: {content: <base64>}, features: [{type: TEXT_DETE..."
}
]
"apis":[1 item
0:"google-cloud-vision-annotate"
]
"specs":[2 items
0:{4 items
"kind":"markdown"
"api":"google-cloud-vision-annotate"
"url":"https://raw.githubusercontent.com/pnyxai/pocket-network-services/6f3912ee520cb75a98288aaf333d6ba1ef729788/mainnet/docs_optical-character-recognition.md"
"sha256":"adadc9550ce472f5e0c460ba7c191f5a199a127c4ee325e2d1a22bc11f4bc14d"
}
1:{4 items
"kind":"google-discovery"
"api":"google-cloud-vision-annotate"
"url":"https://vision.googleapis.com/$discovery/rest?version=v1"
"notes":"Google publishes no OpenAPI document for Cloud Vision; this is its machine-readable Discovery document, living (no sha256). Only images:annotate with TEXT_DETEC..."
}
]
"access":"public"
"results":"variable"
"serving":{3 items
"backend":"Any Google Cloud Vision-compatible server exposing POST /v1/images:annotate (REST) and GET /health. Recommended: image-inference-server (Rust gateway + NVIDIA T..."
"healthcheck":[2 items
0:{4 items
"rpc_type":"REST"
"request":{2 items
"path":"/health"
"method":"GET"
}
"expect":{2 items
"json_path":"$.endpoints.ocr.status"
"matches":"^healthy$"
}
"notes":"Pins the ocr endpoint: /health reports only enabled endpoints, so a detect-only deployment fails it."
}
1:{4 items
"rpc_type":"REST"
"request":{3 items
"path":"/v1/images:annotate"
"method":"POST"
"body":{1 item
"requests":[1 item
0:{2 items
"image":{1 item
"content":"iVBORw0KGgoAAAANSUhEUgAAAUAAAABQAQAAAABASpEoAAABz0lEQVR42u2WMW7cQAxF38wKu+p2yiCN5gAp9gSRDhbEcxQfI+UcYY/AdC5pIIUQjIYpJBuOdyWnDSDWX4/k8JOQM/4tPLtwF+7CN8LJAz/jhtJM..."
}
"features":[1 item
0:{1 item
"type":"TEXT_DETECTION"
}
]
}
]
}
}
"expect":{2 items
"json_path":"$.responses[0].textAnnotations[0].description"
"matches":"(?i)pocket"
}
"notes":"Functional probe: a 320x80 PNG reading POCKET. An image without text yields {}, so a probe without text proves nothing."
}
]
"notes":"Use a server with dynamic batching so one backend serves many concurrent requests. Recommended model: PP-OCRv6 medium det + rec (Apache-2.0); PP-OCRv6 small on ..."
}
"updated":"2026-10-19"
}