I used the features of Accurate-Interpretable-VAD](https://github.com/talreiss/Accurate-Interpretable-VAD).
However, I can't put the velocity, deep_feature and pose together since the length of pose is far less than the deep_feature.
Can u provide the code of dataloader or the way u deal this problem. And btw, on avenue dataset, the model can't perform as well as ur paper with velocity and deep_feature in supplementary material.
I used the features of Accurate-Interpretable-VAD](https://github.com/talreiss/Accurate-Interpretable-VAD). However, I can't put the velocity, deep_feature and pose together since the length of pose is far less than the deep_feature. Can u provide the code of dataloader or the way u deal this problem. And btw, on avenue dataset, the model can't perform as well as ur paper with velocity and deep_feature in supplementary material.