OCR Demos

Compare different OCR engines side-by-side. From browser-based WASM to server-side C++ — test accuracy, speed, and confidence across approaches.

Back to jeffbrand.ca

Individual Engines

Foxit OCR C++ Native

Server-side Foxit PDF SDK compiled to native C++. Fastest engine with built-in deskew, despeckle, and high-precision text extraction.

Engine: Foxit PDF SDK (C++) · Server-side

Foxit OCR LLM Engine Beta 0.3

Server-side Foxit OCR powered by an LLM engine via the Java SDK. Best accuracy on complex layouts. Performance limited without GPU — NVIDIA acceleration planned. Beta 0.3 — not publicly available.

Engine: Foxit LLM OCR (Java) · Server-side · CPU only (no GPU)

Tesseract OCR Server

Tesseract 5 LSTM engine running server-side at 300 DPI with tessdata_best models. Parallel page processing on multiple cores.

Engine: Tesseract 5.3 (Python) · Server-side

Tesseract OCR WASM

Tesseract compiled to WebAssembly via tesseract.js. Runs entirely in your browser — no server needed. Slower but fully private.

Engine: tesseract.js (WASM) · Client-side

Tools

OCR Compare All Engines

Upload one document and run all four OCR engines simultaneously. Compare performance, accuracy, and confidence side-by-side.

Benchmarks all engines · Real-time comparison