/CNN-Tracker-v2

A combination of YOLO and py-MDNet in order to classify and automatically track objects in a real word video sequence

Primary LanguagePythonOtherNOASSERTION

CNN-Tracker-v2

The (just slightly) better version of CNN-Tracker-v1 (which will also most likely be finished)

Acknowledgement

This project relies on different frameworks, libraries, GitHub Projects and user contributions.

  1. py-MDNet
  2. YOLO (darknet)

Thank you everyone for contributing and for all the work that has been put into other projects that we are able to use!

Usage

I will extend this part as soon as I've built a first proper system, that works. If you want to contribute and use this system: so far, this is (not even) a Frankenstein consisting of YOLO and MDNet. Try to wrap your head around it.

First results

13/09/17 - Ability to restart tracking with the same network after completely loosing the object (full occlusion), built the emergency mode afterwards. This is probably one of the most important results so far.