This repository contains the official Pytorch implementation of Temporal Shift GAN for Large Scale Video Generation. Further details can be found here: https://arxiv.org/abs/2004.01823

To reproduce results download the pre-trained weights, then start training.

Pre-trained ImageNet weights: https://drive.google.com/open?id=1QvbsuhVB036wbtc4pKJRHhli0q_eOUGn

To use any dataset you should extract the frames of all videos and create a .txt file with three columns in this exact order: Location of the frames of the video, Number of frames in the folder and Class it belongs to.

Run the script run_video_gan.sh.

The script run create_dataset.sh will automatically create N videos of each class on the dataset.

Download the dataset right here: https://drive.google.com/file/d/15KqP9uxsJ9cMUY5WZyVvl2NlIQe-HQXz/view?usp=sharing