Video-based-Person-Re-identification-Using-Spatial-Temporal-Attention-Networks

Prerequisite downloads -

anaconda - https://askubuntu.com/questions/505919/how-to-install-anaconda-on-ubuntu

pytorch - conda install pytorch torchvision -c pytorch

After that :

Go inside ilids folder and run main.py

Arxiv link of technical report - https://arxiv.org/abs/1810.11261

Please cite this paper if you find this work useful

###Update ! Please check the Non-Local Attention network repository from my profile for updated attention mechanism. This method gives much better results.

Running opticalflow.m will generate the optical flow for your data.