It is an OCR example from Baidu Paddle-Lite used vcam which is run on Raspberry Pi 4 only.
| Directory | Content |
|---|---|
| mainApp | Source code to display AI results. |
| inference_lite_lib.armlinux.armv7hf | Source code to process AI results. |
| bin | Example to run OCR model. |
| public | Library of source code. |
- Step mainApp
- Install opencv 3
sudo apt-get install libopencv-dev
- Install libusb1.0
sudo apt-get install libusb-1.0-0-dev
- Compile in
vcam_ocr/mainApp
sudo make clean
sudo make
- Run in
vcam_ocr/mainApp/bin
sudo ./mainApp
- Step Paddle CRNN Text Recognition Algorithm
cd vcam_ocr/inference_lite_lib.armlinux.armv7hf/demo/cxx/ocr`
sudo make
cd vcam_ocr/inference_lite_lib.armlinux.armv7hf/demo/cxx/ocr/debug
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib/.../ncc_ocr/inference_lite_lib.armlinux.armv7hf/demo/cxx/ocr/debug
./ocr_db_crnn
- Run Bin
cd vcam_ocr/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib/.../bin
./run.sh
- Train model
Training tool running on x86 system for paddle OCR models may help to train model by yourself. Please refer to PaddleOCR Model training/evaluation GitHub for specific usage method.