Provides a PyTorch implementation of OCR (Optical Character Recognition) models, primarily converting models from PaddleOCR to PyTorch format. Includes support for text detection and recognition algorithms with training, evaluation, inference, and model export capabilities. Covers PP-OCR series models including v3 and v4 variants.