GitLab

refactor(ocr): remove redundant code and improve code quality 

refactor(ocr): remove redundant code and improve code quality

- Remove unnecessary GPU checks and cuda() calls
- Consolidate tensor device placement using .to(self.device)
- Add warning suppression for cleaner output
- Refactor conditional logic for better readability

feat: add batch example

refactor(demo): simplify batch_demo.py and update demo.py

- Remove unnecessary imports and code in batch_demo.py
- Update demo.py to use relative paths and improve code structure
- Adjust output directory structure in both scripts
- Remove redundant code and simplify functions

feat: add batch example

update released note

- Add RapidOCR and PaddleOCR2Pytorch to the Acknowledgments list in README.md
- Add RapidOCR and PaddleOCR2Pytorch to the Acknowledgments list in README_zh-CN.md

- Installation and compatibility optimizations:
- Replace PaddleOCR with paddleocr2torch to resolve conflicts between Paddle and PyTorch
  - Remove layoutlmv3 usage to solve compatibility issues with detectron2
  - Extend PyTorch version compatibility to2.2~2.6  - Extend CUDA compatibility to 11.8~12.6
  - Extend Python version compatibility to 3.10~3.12

- Performance optimizations:
 - Support batch processing for multiple PDF files
  - Optimize mfr model loading and usage to reduce memory consumption and improve speed
  - Reduce minimum memory requirement to 6GB
  - Improve running speed on MPS devices

- Parsing effect optimization:
  - Update mfr model to unimernet(2503) to fix line break issues in multi-line formulas

update docs

- Update CUDA version requirements to12.4
- Recommend nvidia-driver-570-server for Ubuntu
- Remove Python version specification for conda environment
- Update magic-pdf version requirement to 1.3.0
- Simplify CUDA acceleration testing instructions
- Remove OCR acceleration with paddlepaddle-gpu
- Update torch and torchvision installation instructions for Windows

- Update Python version requirement to 3.10-3.12
- Expand CUDA environment options to 11.8/12.4/12.6
- Update GPU VRAM requirement to 6GB or more
-

feat: support 3.10~3.12 & remove paddle

- Update matplotlib minimum version to 3.10 for Linux and MacOS
- Specify version ranges for PyYAML, ftfy, openai, shapely, pyclipper, and omegaconf
- Update dill to version <1 for compatibility

- Update PyMuPDF to version <1.25.0
- Update pydantic to version <2.11
- Update transformers to version < 5.0.0
- Remove always_apply parameter from alb.ToGray in image processing

- Update the default configuration path in pytorchocr_utility.py
- Add required dependencies for paddleocr2pytorch in setup.py:
  - shapely
  - pyclipper
  - omegaconf

- Remove unused UniMERNet and LayoutLMv3 model configurations
- Update OCR model path and dictionary path for PaddleOCR
- Modify README to update system requirements and installation instructions
- Update setup.py to include new package data

- Remove unused imports for concurrent.futures, multiprocessing, and paddle
- Delete commented-out code
- Update numpy dependency to remove upper version limit
- Remove InferenceResult import that was commented out

- Remove paddleocr, paddlepaddle, rapidocr-paddle, and rapidocr-onnxruntime from requirements.txt files
- Simplify pip install commands in Dockerfiles
- Remove installation of paddlepaddle-gpu in china and global Dockerfiles
- Update requirements.txt files across all Docker configurations

- Commented out the code that copies the paddleocr model to user directory
- This change affects both download_models.py and download_models_hf.py scripts

- Update download_models.py and download_models_hf.py scripts
- Change OCR model path from paddleocr to paddleocr_torch

- Add newline at the beginning of arabic_dict.txt
- Change mode of multiple dictionary files

- Remove OCR utils, modified PaddleOCR, and StructEqTable model
- Delete related import statements and model definitions
- Update dependencies in setup.py to remove paddlepaddle and related OCR packages

- Comment out print statements in base_ocr_v20.py and pytorch_paddle.py
- Update table model initialization to use lang parameter instead of ocr_engine
- Remove unused RapidOCR initialization in rapid_table.py

- Comment out OCR model initialization and execution for low-contrast spans
- Add batch OCR processing for collected image spans
- Adjust contrast threshold for OCR processing
- Remove unnecessary OCR processing for high-contrast spans
- Implement more efficient OCR workflow by processing multiple spans at once

- Enhance the logging of execution times by adding more detailed function identification
- Implement class name and module name inclusion for better traceability

- Remove unused OCR dictionaries for Arabic, Belarusian, Bulgarian and Armenian languages
- Update model configurations in arch_config.yaml:
- Comment out 'out_channels' for various language models
  - Rename Arabic, Korean, Japanese, Tamil and Devanagari model configurations to use 'v3' instead of 'v4'
- Delete ar_dict.txt, be_dict.txt and bg_dict.txt files
- Update arabic_dict.txt to remove blank line at the start

- Remove unused imports and code
- Simplify model architecture by removing unnecessary components
- Update initialization and forward pass logic
- Rename variables for consistency

- Added warnings module to import list
- Implemented a warning catcher to ignore FutureWarning from the transformers module
- This change prevents unnecessary warning messages during model inference

- Replace direct OCR model access with AtomModelSingleton for better model management
- Round OCR scores to 2 decimal places for consistency
- Improve error handling and logging in batch analysis
- Simplify OCR result processing in pdf_parse_union_core_v2.py

- Add support for multiple languages in OCR processing
- Create separate lists for each language to improve processing efficiency
- Update OCR model initialization to use PytorchPaddleOCR instead of ModifiedPaddleOCR
- Modify get_ocr_result_list function to include language information- Improve logging for OCR detection and recognition

- Split OCR process into detection and recognition stages
- Update batch analysis and document analysis pipelines
- Modify OCR result formatting and handling
- Remove unused imports and optimize code structure

- Add base model structure for OCR in pytorch
- Implement data augmentation and transformation modules
- Create utilities for dictionary handling and state dict conversion
- Include post-processing modules for OCR
- Add weight initialization and loading functions

feat: remove old inference code