Strengthening Skeletal Action Recognizers via Leveraging Temporal Patterns
Abstract
Skeleton sequences are compact and lightweight. Numerous skeleton-based action recognizers have been proposed to classify human behaviors. In this work, we aim to incorporate components that are compatible with existing models and further improve their accuracy. To this end, we design two temporal accessories: discrete cosine encoding (DCE) and chronological loss (CRL). DCE facilitates models to analyze motion patterns from the frequency domain and meanwhile alleviates the influence of signal noise. CRL guides networks to explicitly capture the sequence’s chronological order. These two components consistently endow many recently-proposed action recognizers with accuracy boosts, achieving new state-of-the-art (SOTA) accuracy on two large datasets.
Cite
Text
Qin et al. "Strengthening Skeletal Action Recognizers via Leveraging Temporal Patterns." European Conference on Computer Vision Workshops, 2022. doi:10.1007/978-3-031-25072-9_39Markdown
[Qin et al. "Strengthening Skeletal Action Recognizers via Leveraging Temporal Patterns." European Conference on Computer Vision Workshops, 2022.](https://mlanthology.org/eccvw/2022/qin2022eccvw-strengthening/) doi:10.1007/978-3-031-25072-9_39BibTeX
@inproceedings{qin2022eccvw-strengthening,
title = {{Strengthening Skeletal Action Recognizers via Leveraging Temporal Patterns}},
author = {Qin, Zhenyue and Ji, Pan and Kim, Dongwoo and Liu, Yang and Anwar, Saeed and Gedeon, Tom},
booktitle = {European Conference on Computer Vision Workshops},
year = {2022},
pages = {577-593},
doi = {10.1007/978-3-031-25072-9_39},
url = {https://mlanthology.org/eccvw/2022/qin2022eccvw-strengthening/}
}