In a nutshell

OTLabels is a workflow for the retraining of AI models to detect road users in videos based on existing open source software packages (CVAT, YOLO v5).

Key features:

  • Annotation of vehicles and persons using existing open source software (CVAT)
  • Retrain existing neural network models with YOLO v5 for individual optimazation of the dectection ability
  • Long-term: Real time detection
  • Runs on any Windows laptop. However, a decent GPU with Cuda support is strongly recommended.

Most of the code describing the workflow will be developed by the OpenTrafficCam contributors (based on python libraries like NumPy, pandas, tkinter and OpenCV).

Found errors? Think you can improve this documentation? Simply click the Edit link at the top of the page, and then the icon on Github to make your changes.