FlowCaps: Optical Flow Estimation with Capsule Networks for Action Recognition
Abstract
Capsule networks (CapsNets) have recently shown promise to excel in most computer vision tasks, especially pertaining to scene understanding. In this paper, we explore CapsNet's capabilities in optical flow estimation, a task at which convolutional neural networks (CNNs) have already outperformed other approaches. We propose a CapsNet-based architecture, termed FlowCaps, which attempts to a) achieve better correspondence matching via finer-grained, motion-specific, and more-interpretable encoding crucial for optical flow estimation, b) perform better-generalizable optical flow estimation, c) utilize lesser ground truth data, and d) significantly reduce the computational complexity in achieving good performance, in comparison to its CNN-counterparts.
Cite
Text
Jayasundara et al. "FlowCaps: Optical Flow Estimation with Capsule Networks for Action Recognition." Winter Conference on Applications of Computer Vision, 2021.Markdown
[Jayasundara et al. "FlowCaps: Optical Flow Estimation with Capsule Networks for Action Recognition." Winter Conference on Applications of Computer Vision, 2021.](https://mlanthology.org/wacv/2021/jayasundara2021wacv-flowcaps/)BibTeX
@inproceedings{jayasundara2021wacv-flowcaps,
title = {{FlowCaps: Optical Flow Estimation with Capsule Networks for Action Recognition}},
author = {Jayasundara, Vinoj and Roy, Debaditya and Fernando, Basura},
booktitle = {Winter Conference on Applications of Computer Vision},
year = {2021},
pages = {3409-3418},
url = {https://mlanthology.org/wacv/2021/jayasundara2021wacv-flowcaps/}
}