frotms/PaddleOCR2Pytorch

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1205 stars · 224 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A PyTorch port of PaddleOCR that lets you run PaddleOCR-trained models (detection, recognition, and document structure parsing) without the PaddlePaddle framework. It includes converted models for text detection (DB, DB++, FCENET), recognition (CRNN, SVTR, PP-OCRv2 through v6), plus PP-StructureV3 document parsing with layout detection, table recognition, and formula recognition.

Use cases

• run PaddleOCR models in PyTorch
• extract text from images with OCR
• convert PaddlePaddle models to PyTorch
• parse document structure from PDFs and images
• recognize tables and formulas in documents
• multilingual text recognition in 50 languages
• detect and read seal/stamp text in documents

When to choose

• you want PaddleOCR accuracy but your stack is PyTorch-only
• you need to avoid installing PaddlePaddle
• you need document structure parsing (layout, tables, formulas) in PyTorch
• you need multilingual OCR with lightweight models

When to avoid

• you are fine using the official PaddleOCR with PaddlePaddle
• you need training/fine-tuning rather than inference
• you need guaranteed parity with upstream PaddleOCR releases

Facets

library · maturity active

ocr machine-learning image-processing computer-vision pdf computer-vision image-processing deep-learning pdf python cross-platform paddleocr pytorch text-detection text-recognition model-conversion document-parsing table-recognition formula-recognition multilingual-ocr db crnn natural-language-processing gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="frotms/PaddleOCR2Pytorch")