Openlearnia

← All tools

Browser OCR

Browser (WebGPU or ONNX/WASM)

OCR suite: Granite-Docling, RapidOCR EN/ZH, and Tesseract — each with its own UI.

Privacy-first browser OCR suite with dedicated routes per engine. Granite-Docling (~1 GB WebGPU) for DocTags/HTML/Markdown/crops; RapidOCR English (~14 MB) and Chinese (~16 MB) via ONNX; Tesseract.js eng (~23 MB) as classic fallback. Opt-in downloads with size disclosure. Images or pdf.js PDF pages (one at a time). Page bytes stay on device.

Open app GitHub
Platform Browser (WebGPU or ONNX/WASM)
Live app https://docling-extract.openlearnia.com
Source https://github.com/openlearnia/docling-extract
Privacy Runs locally — nothing leaves your device

Features

  • Suite hub + dedicated UI per model (not a buried picker)
  • Granite-Docling-258M ONNX via Transformers.js + WebGPU
  • RapidOCR English + Chinese PP-OCRv4 via ONNX Runtime Web
  • Tesseract.js eng WASM fallback
  • Planned stubs: PaddleOCR.js, Florence-2, LightOnOCR, TrOCR, GLM-OCR