Extract GTINs from a product photo via Bedrock Vision

Server-side fallback for in-browser Tesseract.js OCR. Uploads a single product photo (JPEG/PNG/GIF/WebP) to Bedrock Vision, asks the vision model for barcode digit sequences, then keeps only candidates that pass the GS1 mod-10 check digit. The response is always shaped — on any backend failure it returns an empty candidates list rather than surfacing the error, so the calling UI can fall through to a manual-entry prompt without bespoke error handling.

POST /dashboard/api/v1/ocr/gtin

Extract GTINs from a product photo via Bedrock Vision

cURL

curl --request POST \
  --url https://www.closient.com/dashboard/api/v1/ocr/gtin \
  --header 'Content-Type: multipart/form-data' \
  --header 'X-API-Key: <api-key>' \
  --form image='@example-file'

200

{
  "candidates": [
    "<string>"
  ]
}

Authorizations

APIKeyHeaderAuthOAuthTokenAuthSessionAuth

X-API-Key · string · header · required

Body

multipart/form-data

image · file · required

Product photo to scan for GTINs (JPEG, PNG, GIF, or WebP).

Response

200 - application/json

OK

Validated GTIN candidates extracted from a product photo.

candidates · string[] · required

Zero or more GTIN candidates that pass the GS1 mod-10 check digit. Strings are zero-padded to 14 digits. Empty when the model returned nothing usable or Bedrock errored — callers should treat empty as 'OCR found no GTIN' rather than a transport failure.

GTIN barcode digits. Accepts 8/12/13/14 digit forms; the resolver and storage layers normalize to GTIN-14 by left-padding with zeros.

Pattern: ^\d{8,14}$

source · enum<string> · required

Backend that produced the candidates. Currently always bedrock; reserved as an enum for future server-side OCR fallbacks.

Available options: bedrock

Example:

"bedrock"