search

yuz1wan / video_distillation

Official implementation of Dancing with Still Images: Video Distillation via Static-Dynamic Disentanglement.
15 stars 2 forks source link

Could you release a distilled dataset? #3

Closed Wu-ssssss closed 1 month ago

Wu-ssssss commented 1 month ago

I am so curious what a video will look like when distilled.

yuz1wan commented 1 month ago

Yes, of course. We can provide some checkpoints of distilled videos.

We have provided a visualized result of DM+Ours HMDB51 spc2 dpc2 at /HMDB51/DM+Ours/ (every row is a synthetic 16-frame video).

We also provide a Jupyter Notebook to convert the checkpoint back into frames.

Wu-ssssss commented 1 month ago

Thanks for your reply!

I have tried the vis.ipynb and some problems occurred. Is the sta.pt omitted? I don't actually know which .pt file corresponds to "path_static".

Yes, of course. We can provide some checkpoints of distilled videos.

We have provided a visualized result of DM+Ours HMDB51 spc2 dpc2 at /HMDB51/DM+Ours/ (every row is a synthetic 16-frame video).

We also provide a Jupyter Notebook to convert the checkpoint back into frames.

yuz1wan commented 1 month ago

Apologies for the delayed response, as I've been dealing with some critical deadlines recently. "sta.pt" refers to our static memory, which you can train yourself. Of course, we also provide our pre-trained static memory (as mentioned in the README file).

Wu-ssssss commented 1 month ago

Apologies for the delayed response, as I've been dealing with some critical deadlines recently. "sta.pt" refers to our static memory, which you can train yourself. Of course, we also provide our pre-trained static memory (as mentioned in the README file).

Thanks a lot! :muscle: