Commit History
-
Initial release: EasyOCR ONNX models with JPQD quantization
- Add CRAFT text detection model (5.7KB, 1.51x compression) - Add English text recognition model (8.5MB, 3.97x compression) - Add Latin text recognition model (8.5MB, 3.97x compression) - Include comprehensive documentation and usage examples - Total size reduction: 108MB → 17MB (6.4x compression) - Full Python implementation with preprocessing/postprocessing - HuggingFace Hub compatible with proper metadatac1ac2fb