GitHub radar
Pull text from scans, photos and PDFs right on your laptop
A small text-recognition model from Baidu can read a whole multi-page document in one pass, with no splitting into chunks. It runs on your own computer, no cloud subscription needed.
It's a program that reads text from photos, scans and PDF files and turns it into plain text you can copy.
When it helps
Useful for a one-off job pulling text out of a scan, a photo of a whiteboard, or a PDF report. For ongoing serious document work, look for something more capable.
Pros
- Runs on an ordinary laptop, no internet or subscription
- MIT license — use it with no restrictions
- Reads a long multi-page document in a single pass
- Compressed builds already exist, around 2GB in size
Cons
- Mid-tier quality — not a replacement for serious document work
- You'll need to pick and set up the right build yourself
How to set it up — step by step
- 1Open your AI agent (Claude Code or similar) in your project folder.
- 2Send the agent the text from the block below — it will install the model and pick the right version for your machine.
- 3Have the agent run it through Ollama or LM Studio.
- 4Feed the model a scan, photo, or PDF and ask it to pull out the text.
- 5Check the result — compare the extracted text against the original for typos.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up the Unlimited-OCR model by Baidu locally for me: https://huggingface.co/baidu/Unlimited-OCR — pick a quantized version for my computer, run it through Ollama or LM Studio, and show me how to feed it a scan or a PDF to pull out the text.
A regular laptop with 8 GB of RAM or more — the compressed version takes about 2 GB.
Open on Hugging Face▌ More finds