Utilities · sync
ocr-vision
Extract text from any image — 100+ languages, no AI keys.
Recognised text
About
ocr-vision wraps Tesseract.js — the WebAssembly port of Google's Tesseract — and exposes it as a one-call API. No native dependencies, no Docker, no language servers; the worker downloads ~10 MB of language data on first use and caches it.
Use it for: receipt parsing, signage transcription, scanned PDFs (one page at a time), handwritten note digitisation (when neat enough).
Language codes follow Tesseract's three-letter convention: 'eng', 'spa', 'fra', 'deu', etc. Combine languages with '+' (e.g. 'eng+jpn') when documents are bilingual. Output includes the full text plus the first 1000 word-level bounding boxes for layout-aware downstream processing.
SSRF guard: requests with imageUrl are rejected when the host resolves to a private / loopback / link-local address. Input is capped at 25 MB.
{
"type": "object",
"oneOf": [
{
"required": [
"imageUrl"
]
},
{
"required": [
"imageBase64"
]
}
],
"properties": {
"imageUrl": {
"type": "string",
"format": "uri",
"description": "https URL to the source image (max 25 MB)."
},
"imageBase64": {
"type": "string",
"description": "Base64-encoded image bytes (raw or data: URL)."
},
"language": {
"type": "string",
"default": "eng",
"description": "Tesseract language code; combine with '+' (e.g. 'eng+fra').",
"pattern": "^[a-z]{3}(\\+[a-z]{3})*$"
}
}
}These are descriptive previews. Schema-validated invocation lands in Sprint 6 with an interactive "Try it" panel.
Reviews (0)
No reviews yet — be the first to share your experience.