DEEP LEARNING FOR VISUAL TRACKING

This notebook aim to gather the contributions of both group working on deep learning approach to visually track an object in a video. Two different framework were used : GOTURN and YOLO.

The architecture used by GOTURN algorithm is described in the original paper :

Learning to Track at 100 FPS with Deep Regression Networks,
David Held, Sebastian Thrun, Silvio Savarese,

The authors has implemented it with caffe : davheld/GOTURN

The tensorflow implementation that has been used to develop our own code: tangyuhao/GOTURN-Tensorflow

Illustration of how this network works:

Repository of the project :

• Link to slides : Slides

---

• GOTURN implemented in a colab ipython Notebook can be found in the following : Code

• YOLO Implemented in C : Chay16/TrackingVideo