open-mmlab/mmocr

Health v2 · maintenance only

Adoption not part of the score

4752 stars · 781 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

MMOCR is OpenMMLab's PyTorch-based toolbox for text detection, recognition, and key information extraction. It provides a model zoo of OCR algorithms (DBNet, PANet, CRNN, ABINet, SVTR, etc.) with training, inference, and deployment utilities.

Use cases

• extract text from images
• detect text regions in photos
• build an ocr pipeline in python
• train a custom text recognition model
• extract key information from scanned documents
• run scene text recognition with pretrained models

When to choose

• you need state-of-the-art text detection or recognition models in PyTorch
• you want to train or fine-tune OCR models on custom datasets
• you need key information extraction from documents
• you are already using the OpenMMLab ecosystem

When to avoid

• you just need a simple off-the-shelf OCR for PDFs without ML setup (e.g. Tesseract)
• you need a lightweight CPU-only production deployment
• you work outside Python/PyTorch

Facets

library · maturity active

ocr machine-learning deep-learning computer-vision nlp computer-vision image-processing deep-learning machine-learning python windows text-detection text-recognition key-information-extraction pytorch openmmlab model-zoo natural-language-processing linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="open-mmlab/mmocr")