Description

We evaluated the AVID-CMA model on HMDB-51 dataset by tuning the training parameters to improve the accuracy slightly. We implemented a combination of Cross-Entropy Loss and MSE-Loss in order to balance the loss function for this particular dataset.

Contributors

Lakshya Gupta
Parshant Arora

Image credits

• Imgur and BeFunky
• https://arxiv.org/pdf/2004.12943.pdf