#019

LLM-aided OCR

github.com/Dicklesworthstone/llm_aided_ocr →

OCRLLM post-processingScanned documentsOpen sourceLocal inference

▶ 591 views♥ 31 likes💬 0 comments🔗 111 visits

Tesseract gets it wrong. Let an LLM fix it

WHAT IT SOLVES

Tesseract OCR is free and everywhere, but it mangles characters constantly. Instead of retraining or tuning, this repo hands the messy output to an LLM and says: proofread this

WHY IT'S INTERESTING

★ Product taste

Don't rebuild the wheel — patch it

Tesseract writes the rough draft, the LLM does copyediting. Scanned PDF → Tesseract raw text → LLM corrects per-chunk and spits out clean Markdown. The clever bit is smart chunking: you can't stuff a whole page into a context window, so it breaks text semantically, fixes each piece, then stitches it back

★ Real craft

Works with local models too

Not locked to a single API — supports both local LLMs and cloud endpoints. For anyone scanning contracts or medical docs, that's not a nice-to-have, it's the whole point. Docker-ready, 63 commits, proper changelog — this isn't a weekend drop-and-forget