We introduce Knowledge Fusion Transformers for video action classification. We present a self-attention based feature enhancer to fuse action knowledge in 3D inception based spatio-temporal context of the video clip intended to be classified. We show, how using only one stream networks and with little or, no pretraining can pave the way for a performance close to the current state-of-the-art. Additionally, we present how different self-attention architectures used at different levels of the network can be blended-in to enhance feature representation. Our architecture is trained and evaluated on UCF-101 and Charades dataset, where it is competitive with the state of the art. It also exceeds by a large gap from single stream networks with no to less pretraining.
@article{arxiv.2009.13782,
title = {Knowledge Fusion Transformers for Video Action Recognition},
author = {Ganesh Samarth and Sheetal Ojha and Nikhil Pareek},
journal= {arXiv preprint arXiv:2009.13782},
year = {2020}
}