Guangming Zhu

I noticed that the class BouncingMNISTDataHandler was not used when training. Can you just share with me your generated moving mnist dataset which can be used in train.py? Thank you!

@peach2wanzi This work is done with the old version of Tensorlayer. You can refer to the new network using keras in TF-1.2: https://github.com/GuangmingZhu/ContinuousGR and https://github.com/GuangmingZhu/AttentioninConvLSTM.

@agunapal Sorry, I can't access the dropbox website.

@shajie17 Link: https://pan.baidu.com/s/1_qlvYWdY-2nW8dJSQWzuoA Code: zjyr

@Lihongjian520 The 20BN-jester Dataset V1 is used under license agreement. Therefore, I can't distribute the dataset without permission. You can download the dataset from the official website https://20bn.com/datasets/jester.

112 is the one average value of three channels (RGB) of videos.

For example, we extracts images from M_person_1_backgroud_1_illumination_1_pose_1_actionType_10.avi, and then store the images in the dir 11119. The "9" is the action type which starts from 0.

@peach2wanzi they are the frame count .

@sumeetssaurav Which version of TF are you using? I'm not sure the compatibility with the newer version than TF-1.2. I have seen the similar error when I was using a...

@sumeetssaurav Besides, the typo "clstm_respahe" does not exist in this code. I am not sure which code you are using.