PaddleOCR tool library v2.9.1

PaddleOCR aims to create a rich, leading, and practical OCR tool library to help users train better models and implement applications.

PP-OCR is a practical ultra-lightweight OCR system. It mainly consists of three parts: DB text detection, detection frame correction and CRNN text recognition. The system adopts 19 effective strategies from 8 aspects: backbone network selection and adjustment, prediction head design, data enhancement, learning rate transformation strategy, regularization parameter selection, pre-training model use, and automatic model cropping and quantification. The model was optimized and slimmed down, and finally an ultra-lightweight Chinese and English OCR with an overall size of 3.5M and an English digital OCR of 2.8M were obtained.

characteristic

1. PPOCR series high-quality pre-training model, accurate recognition effect

Ultra-light ppocr_mobile mobile series: detection (2.6M) + direction classifier (0.9M) + recognition (4.6M) = 8.1M

General ppocr_server series: detection (47.2M) + direction classifier (0.9M) + recognition (107M) = 155.1M

Ultra-lightweight compression ppocr_mobile_slim series: detection (1.4M) + direction classifier (0.5M) + recognition (1.6M) = 3.5M

2. Support Chinese and English number combination recognition, vertical text recognition, and long text recognition

3. Support multi-language recognition: Korean, Japanese, German, French

4. Support user-defined training and provide rich predictive inference deployment solutions

5. Support PIP quick installation and use

6. Can run on Linux, Windows, MacOS and other systems