Skip to content

Latest commit

 

History

History
36 lines (30 loc) · 1.09 KB

README.md

File metadata and controls

36 lines (30 loc) · 1.09 KB
Raw

Yolo Object Detector

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

image

NOTE Due to file size restrictions here in Github I have not included the ONNX models in the repo.

Build and run

  • 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

Tasks

  • 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