Defect detection with unet model

Training Unet model base on resnet50 and apply cpp for deployment base on ggml repository

Quick start

Download files: modelunet.gguf, ggml.dll and unet.exe

cd unet folder
unet -i image.jpg

Running with CUDA Download ggml.dll for CUDA

cd unet folder
unet -i image.jpg

or pull repository ggml and build on window os

cd ggml
cmake -DGGML_CUDA=ON -DCMAKE_CUDA_COMPILER="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.6\bin\nvcc.exe
msbuild ALL_BUILD.vcxproj /p:Configuration=Release 

Convert file h5 model to gguf

Request tensorflow 2.15, download file h5 model

python convert.py modelunet.h5

Training model

Training file Unet_detection.ipynb, download data set here

Run speed

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

MIT

References

• ggml
• MiAI_Defect_Detection