Skip to main content

OcrProvider

@forge/monorepo


@forge/monorepo / backend/src / OcrProvider

Interface: OcrProvider

Defined in: backend/src/documents/vision.ts:79

Transcribes an image or a scanned document.

Takes the original bytes and media type, not page images: rasterising a PDF needs a renderer, and every real OCR service (Textract, Document AI, Azure Document Intelligence) accepts a PDF directly and does that itself. Making rasterisation the provider's business is both simpler and what the APIs already assume.

Properties

id

readonly id: string

Defined in: backend/src/documents/vision.ts:80


mediaTypes

readonly mediaTypes: readonly string[]

Defined in: backend/src/documents/vision.ts:81

Methods

recognise()

recognise(input): Promise<ExtractionFailure | OcrResult>

Defined in: backend/src/documents/vision.ts:82

Parameters

input
bytes

Uint8Array

limits

ExtractionLimits

mediaType

string

Returns

Promise<ExtractionFailure | OcrResult>