PY-GOTURN
News
I'm back on working on this codebase again, can expect the following changes on the development branch goturn-dev
-
Code clean up
-
Bug fixes, if any
-
Try to add much better interface for easy debugging - Visualization of training, understanding different modules of the code.
-
Documentation or blog on the code.
NOTE: Please switch to goturn-0.1 branch to try out stable version.
This is the python implementation of GOTURN: Generic Object Tracking Using Regression Networks.
Learning to Track at 100 FPS with Deep Regression Networks,
David Held,
Sebastian Thrun,
Silvio Savarese,
Outputs
| Car | Sunshade |
|---|---|
 |
 |
Why implementation in python, when C++ code is already available?
- Easy to understand the overall pipeline of the algorithm in detail
- Easy to experiment new ideas
- Easy to debug and visualize the network with tools like visdom
- Little effort in portability to other OS
Functionalites added so far
-
Training the deep network on Imagenet and ALOV dataset
-
Test code for VOT
How to train your own tracker
To train your own tracker, please follow the guide
How to test on VOT dataset
To test your own tracker on VOT2014, please follow the guide
Contact
Please write to me at nrupatunga.tunga@gmail.com, if you have any suggestions or any new functionality you like to see.