This license plate detector is a technology demonstration of how to use ONNX under Linux.
I annotated license plate images (4000+ jpeg files) and trained my model using YOLOv8 YOLOv26. It is quite convenient to use Yolo26 since I don't have to run Non-Maximum Supression.
- Long plates
- Square(ish) plates
- Foreign plates
In order to load and use ONNX runtime under Linux:
-
Download ONNX Runtime from: https://www.nuget.org/api/v2/package/Microsoft.ML.OnnxRuntime
-
Extract it (it is basically a zip file) to a folder
-
Go to
runtimes/linux-x64/native -
Copy
libonnxruntime.soto./onnx/libfolder of your project -
ln -s libonnxruntime.so libonnxruntime.so.1
-
Copy include files (recursively) to your project folder as well
cp -r build/native/include/ ./onnx/
sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2