A PyTorch implementation of TMU [paper], a differentiable framework named transferable memory, which adaptively distills knowledge from a bank of memory states of multiple pretrained RNNs, and applies it to the target network via a novel recurrent structure called the Transferable Memory Unit (TMU).
Video prediction networks have been used for precipitation nowcasting, early activity recognition, physical scene understanding, model-based visual planning, and unsupervised representation learning of video data.
-
Install Python 3.7, PyTorch 1.3, and OpenCV 3.4.
-
Download data. This repo contains code for two datasets: the Moving Mnist dataset and the KTH action dataset.
-
Train the model. You can use the following bash script to train the model. The learned model will be saved in the
--save_dirfolder. The generated future frames will be saved in the--gen_frm_dirfolder.
cd script/
sh tmu.sh
PredRNN++: Towards A Resolution of the Deep-in-Time Dilemma in Spatiotemporal Predictive Learning.
Yunbo Wang, Zhifeng Gao, Mingsheng Long, Jianmin Wang, and Philip S. Yu.
ICML 2018 [paper] [code]
You may send email to [email protected], [email protected] or [email protected].