Presentation | 2021-12-16 Supervoxel-based Explanation for Action Recognition Ying Ji, Yu Wang, Kensaku Mori, Jien Kato, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Deep neural network has shown remarkable performance in various areas, including image classification, action recognition, and language processing. Despite its popularity, the decision procedure of network still lacks transparency and interpretability, making it difficult to improve the performance further. Recent research in explainable artificial intelligence has explored the relationship between image pixels and output predictions. However, due to the computation cost and complexity of video data, the explanation for video analysis remains unsolved. In this work, we propose a novel supervoxel-based explanation (SVE) method for action recognition. Our approach has two main advantages: (1) videos are represented by concept-level supervoxels, which is easy for human to understand; (2) a concept importance explanation (CIE) framework can estimate the importance rank of different voxels. The experiment results on two kinds of 3d neural networks demonstrate that SVE can explore the most meaningful concepts in a video while predicting. We also visually show the influence of different concepts which is easy to understand. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | explainable artificial intelligenceaction recognitionvideo analysis |
Paper # | PRMU2021-42 |
Date of Issue | 2021-12-09 (PRMU) |
Conference Information | |
Committee | PRMU |
---|---|
Conference Date | 2021/12/16(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Seiichi Uchida(Kyushu Univ.) |
Vice Chair | Masakazu Iwamura(Osaka Pref. Univ.) / Mitsuru Anpai(Denso IT Lab.) |
Secretary | Masakazu Iwamura(NTT) / Mitsuru Anpai(Tottori Univ.) |
Assistant | Kouta Yamaguchi(CyberAgent) / Yusuke Matsui(Univ. of Tokyo) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding |
---|---|
Language | ENG |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Supervoxel-based Explanation for Action Recognition |
Sub Title (in English) | |
Keyword(1) | explainable artificial intelligenceaction recognitionvideo analysis |
1st Author's Name | Ying Ji |
1st Author's Affiliation | Nagoya University(Nagoya Univ.) |
2nd Author's Name | Yu Wang |
2nd Author's Affiliation | Ritsumeikan University(Ritsumeikan Univ.) |
3rd Author's Name | Kensaku Mori |
3rd Author's Affiliation | Nagoya University(Nagoya Univ.) |
4th Author's Name | Jien Kato |
4th Author's Affiliation | Ritsumeikan University(Ritsumeikan Univ.) |
Date | 2021-12-16 |
Paper # | PRMU2021-42 |
Volume (vol) | vol.121 |
Number (no) | PRMU-304 |
Page | pp.pp.98-100(PRMU), |
#Pages | 3 |
Date of Issue | 2021-12-09 (PRMU) |