Michael_Hsu
Michael_Hsu
My mistake, you are right, one should be latent_feature. Thanks!
Hi. 1. Yes we have to preprocess the KTH video by slicing each video into frames. 2. About the code, I think the problem will be how you arrange your...
Yes, we need ground truth while training. Since, for example, 5 image inputs with 5 image predictions, the code will load 10 images at the start of each loop, and...