Pengxiang Li
Pengxiang Li
In my opinion, there should be no such error, can you make a breakpoint here to see what `keys` are in `img_meta`?
Hi, The dataset of youtubevis does not support local validation. You can only generate zip file and submit it to codalab for validation. The metric code is only reserved for...
Yes, it seems that we need to train on video to achieve better results. In short, if LLM can understand the behavior of a time series, I think it would...
use this "stabilityai/stable-video-diffusion-img2vid"
I apologize for the confusion. The situation is such that "mini" simply refers to the 5 videos I personally selected from the bdd tracking dataset for overfitting experiments. Therefore, you...
Hello, unfortunately, the current code does not support text2video, it only supports img2video. If you want to FT img2video, you can refer to the readme.
Yes, I overlooked this aspect, the normalization for images in my dataloader is incorrect, I will fix this.
Fixed. 5d155c1326f1a5db45618054f222c867c59067ef
Thank you for the great suggestions! I really appreciate you sharing these tips. I will handle these changes in the near future, and if you are interested, you are also...
> Can I help with this? What's necessary to finish it? Thank you very much for your attention~ At that time, I was limited by the lack of large-scale video-text...