konstantindob / ocr

OCR pipeline in PyTorch

OCR

OCR pipeline detects and recognizes text in images. OCR was written in PyTorch and allows use ONNX inference and OpenVINO optimization. Detailed info about the neural networks used can be found here: recognition model, detection model.

Installation

Requirements:

• numpy==1.26.2
• opencv-python==4.5.5.62
• PyYAML==6.0
• torch==1.13.0
• onnx==1.12.0
• onnxruntime==1.11.1
• openvino==2023.1.0
• git+https://github.com/KonstantinDob/ocr_detection
• git+https://github.com/KonstantinDob/ocr_recognition

From pip:

python -m build
pip install dist/ocr-1.0.0-py3-none-any.whl

From source:

pip install -U git+https://github.com/KonstantinDob/ocr

Run Inference

To run Inference you should prepare weights and configs. After python3 bin/inference.py --datapath=YOUR_PATH_TO_DATA If you want to use ONNX inference or OpenVINO optimization you should use scripts from ocr/modules.

Docker

To use docker with GPU you need nvidia-docker == 2.9.0.

Build project:

make build

Run in interactive mode:

make run