OpenOCR

Run `npx skills add Topdu/OpenOCR --all -g -y` in your terminal to install all tools in this suite globally.

OpenOCR detects and recognizes text in any image with a single command, returning the text content plus confidence scores. It works on photos, screenshots, and scanned pages.

Yes. Its document parsing mode analyzes page layout, recognizes text, formulas, and tables, then exports clean Markdown or JSON files for each page.

Yes. The UniRec model recognizes formulas as LaTeX and extracts table structures, handling mixed documents with text, equations, and grids in one pass.

Yes. The models use only 0.1B parameters and run on CPU via ONNX, so no GPU or cloud service is required, though a GPU speeds up large batches.