ryfan-rs/video-prediction

spatial-temporal prediction for video

We implemented causualLSTMnet and convLSTMnet. In main_gan.py, we apply convLSTMnet. disNet.py is the code for discriminator.

The model we train on convLSTM only is stored in model_best.pth. The latest model we train on GAN combined convLSTM and changed ResNet is stored in model_generator_new_pretrain.pth.

If you want to test the model, you should run command below python main_gan.py --test --save_g

If you want to train the model from the begining, just run python main_gan.py --pretrain

--pretrain is an optional argument depending on that you choose a pretrained model or not. The default pretrained model is model_best.pth.

Other more optional arguments can be gained by run python main_gan.py --help