l-yezhu / d2m-gan Goto Github PK

Could you share the data process code about original video?

Hello, how do you extract the motion features and visual features of the video, do you have a specific reference link? Looking forward to your reply

Hello, is there any public code for this part of the Genre and Diversity evaluation index?

Hello, if my audio and video are all 10 second files and I want to generate 10 second files, what should I do

Hi, I'd like to report a bug (maybe) in l.96, class VAMDataset, d2m/dataset.py, where the directories of motion files are not post-processed like the audio/video files above. Adding self.motion_files = [Path(motion_files).parent / x for x in self.motion_files] could be a solution?

I am wondering why the motion is in shape of (120*219) for the preprocessed data. Where did 219 come from ?

What should I do if I want to use my own audio/video moves, etc.?

Hi author, Nice Work!
I have tried the generation process on AIST++, it seems that the generated samples have the basic rhythm but don't sound real enough. I don't know what's the problem with it. Is it the reason for high level model setting? And I wonder if you could share the pre-trained models of low level for quick start?

Hi，Is there a checkpoint trained on the tiktok dataset

While reading the paper, in page2 the section with regrard to the hinge loss function, I think you should use max() instead min()

FileNotFoundError: [Errno 2] No such file or directory: 'dataset/aistplus/audio_s2/test/gPO_sBM_c01_d11_mPO1_ch02_seg1.wav'