To understand the world, we humans constantly need to relate the present to the past, and put events in context. In this paper, we enable existing video models to do the same. We propose a long-term feature bank---supportive information extracted over the entire span of a video---to augment state-of-the-art video models that otherwise would only view short clips of 2-5 seconds. Our experiments demonstrate that augmenting 3D convolutional networks with a long-term feature bank yields state-of-the-art results on three challenging video datasets: AVA, EPIC-Kitchens, and Charades.
@article{arxiv.1812.05038,
title = {Long-Term Feature Banks for Detailed Video Understanding},
author = {Chao-Yuan Wu and Christoph Feichtenhofer and Haoqi Fan and Kaiming He and Philipp Krähenbühl and Ross Girshick},
journal= {arXiv preprint arXiv:1812.05038},
year = {2019}
}
Comments
Code and models are available at https://github.com/facebookresearch/video-long-term-feature-banks