thuml/predrnn-pytorch

Do you have the kth_action dataset?

toddwyl opened this issue · 9 comments

Do you have the kth_action dataset ?
I use the kth_action dataset from the official website, and preprocess the video to frames.
But I cannot get the training set of 108,717 and test set of 4,086 sequences as the paper mention.
I get like below:

training set:
there are 183861 pictures
there are 30379 sequences
test set:
there are 105855 pictures
there are 17145 sequences

Can u give me the preprocess method or the preprocessed dataset ?

@wyb15 And I found the moving mnist dataset have no MNIST-3 dataset, could u offer me the dataset?
I could not get the kth test sequences until now. But I get the similar training set which have 115024 sequences. Could u offer me the dataset kth and MNIST-3?
Thanks a lot!

Hi,
were you able to train it on KTH data successfully? I tried but I am not able to get the results which are reported in the paper.

@GuptaSonam I cannot get the correct data and cannot reproduce the results.

@wyb15 Would it be possible for you to share the preprocessed KTH data? I trimmed the videos manually but the results that I am getting are far from the values reported in the paper.
Thanks

@ManWingloeng I've got the KTH dataset. I can send it to you, please let me know your email address.
Btw, is the training code incompletely for training KTH dataset?

@ml18j47z : Thanks. My email id is cs18d005@cse.iitm.ac.in I would be glad to get the processed KTH data.
I wrote the training script myself similar to the mnist training script provided in the repository. I can share it with you if you want.

@GuptaSonam I have sent the dataset to you. Please check it. Can you share this part of the code with me? My email is ml18j47z@leeds.ac.uk.Thanks a lot!

@ml18j47z My email id is yongliangwen@foxmail.com . I use the code e3d_lstm/src/data_provider/kth_action.py in data_provider, but you should know e3d_lstm have some bugs in e3d_lstm model.
Thanks a lot.

Hi, we upload the dataset to Tsinghua cloud, you can find here:
https://cloud.tsinghua.edu.cn/d/7d19372a621a4952b738/