* feat(python): support PP-OCRv6 for onnxruntime and openvino
Add PP-OCRv6 rec and det models (tiny/small/medium) for onnxruntime and openvino engines. v6 ships no cls model; reuse PP-OCRv5 cls.
- typings.py: add OCRVersion.PPOCRV6 and ModelType.TINY/SMALL/MEDIUM
- default_models.yaml: add PP-OCRv6 blocks under onnxruntime/openvino with PaddlePaddle official ModelScope URLs, verified SHA256, and dict_url (PaddleOCR dict files)
Verified end-to-end on ch_en_num.jpg with v6 small det+rec + v5 cls (onnxruntime): 18 text lines recognized correctly.
* fix(python): use uniquely-named PP-OCRv6 model URLs to avoid cache collision
PP-OCRv6 official ONNX exports are all named inference.onnx, causing det and rec models to overwrite each other in RapidOCR's cache (cache_dir/Path(url).name). Switch model_dir/dict_url to a mirror with unique filenames (ch_PP-OCRv6_<task>_<size>.onnx) under modelscope.cn/models/jaminmei/PP-OCRv6-onnx@v1.0. SHA256 unchanged (byte-identical files, renamed only).
Verified onnxruntime + openvino: no more redownload-on-collision; second run is pure cache hit.
* test(python): add PP-OCRv6 det/rec unit tests for onnxruntime and openvino
Mirror the test_ppocrv5_det/rec_mobile pattern. v6 small det on ch_en_num.jpg -> 18 boxes; v6 small rec on text_rec.jpg -> '韩国小馆'. Parametrized over the two engines v6 supports (ONNXRUNTIME, OPENVINO). Existing 100 tests unaffected (verified via collect-only).
Verified: 4 new tests pass (onnxruntime + openvino x det/rec).
* chore(python): switch PP-OCRv6 model URLs to official RapidAI/RapidOCR repo
- det/rec x tiny/small/medium now point to modelscope.cn/RapidAI/RapidOCR (official SWHL conversion, not contributor mirror)
- update SHA256 to the official files' hashes
- drop rec dict_url: official ONNX embeds the character list in metadata (session.have_key() path), matching v4/v5 onnx convention
- verified: 4 v6 unit tests pass; det benchmark Hmean identical to prior mirror-based run (0.8263/0.8492/0.8421 for tiny/small/medium)
* feat(inference_engine): add TensorRT engine support
- Add TensorRT inference session with dynamic shape support
- Implement automatic engine building and caching per GPU architecture
- Add pre-allocated buffers with pinned memory optimization
- Support FP16/FP32/INT8 precision modes
- Add optimization profiles for det/rec/cls models
- Include comprehensive error handling and resource cleanup
- Add unit tests for TensorRT engine (skipped when unavailable)
- Add TensorRT configuration section in config.yaml
* fix(tensorrt): dynamically allocate output buffers to prevent overflow
* Update test_engine.py with TensorRT support final version
* feat(tensorrt): add model_type to engine cache key
- Include model_type in TensorRT engine cache key generation
- Ensures proper cache differentiation based on model_type
* fix(tensorrt): update maximum shape configurations for input and output tensors
- Increased max_shape values in config.yaml and engine_builder.py to 2048 for both detection and recognition tasks.
- Updated default maximum shapes in memory_utils.py to reflect the new limits for input and output buffers.
* feat(tensorrt): implement square input handling for MULTI models
- Added functionality to check if the model requires square input based on configuration.
- Implemented padding of input to square shape and cropping of output back to original dimensions.
- Introduced methods to determine maximum profile size and handle square input requirements for TensorRT inference.
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
SWHL
JinmingMei
locviettriandCursor