This project contains a .NET library and application to detect object in an RTSP video stream using a Yolo model, ML.NET and OpenCV. This is a work in progress.
- The Yolov4 implementation works
- The YoloV2 implementation in incomplete and does not yet work
NOTE Due to file size restrictions here in Github I have not included the ONNX models in the repo.
- Using VS 2022
- Clone the repo
- Download the Yolov4.onnx model
- Build the solution
- Copy the model to the build output folder of the application
- Set the rtsp url including logon credentials as run argument
- Run the application
- Application
- Initial app connect to rtsp
- Propper parameter parsing
- Add ui
- Library
- Yolo support
- Unify code for algoritms
- Config for object clases
- Yolov4 support
- Initial prototype
- TinyYolov2 support
- Initial prototype
- Other Yolo versions TBD
- Yolo support