Human action recognition by SOM considering the probability of spatio-temporal features

研究成果: 著書/レポートタイプへの貢献会議での発言

3 引用 (Scopus)

抄録

In this paper, an action recognition system was invented by proposing a compact 3D descriptor to represent action information, and employing self-organizing map (SOM) to learn and recognize actions. Histogram Of Gradient 3D (HOG3D) performed better among currently used descriptors for action recognition. However, the calculation of the descriptor is quite complex. Furthermore, it used a vector with 960 elements to describe one interest point. Therefore, we proposed a compact descriptor, which shortened the support region of interest points, combined symmetric bins after orientation quantization. In addition, the top value bin of quantized vector was kept instead of setting threshold experimentally. Comparing with HOG3D, our descriptor used 80 bins to describe a point, which reduced much computation complexity. The compact descriptor was used to learn and recognize actions considering the probability of local features in SOM, and the results showed that our system outperformed others both on KTH and Hollywood datasets.

元の言語英語
ホスト出版物のタイトルNeural Information Processing
ホスト出版物のサブタイトルModels and Applications - 17th International Conference, ICONIP 2010, Proceedings
ページ391-398
ページ数8
エディションPART 2
DOI
出版物ステータス出版済み - 12 21 2010
イベント17th International Conference on Neural Information Processing, ICONIP 2010 - Sydney, NSW, オーストラリア
継続期間: 11 22 201011 25 2010

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
番号PART 2
6444 LNCS
ISSN(印刷物)0302-9743
ISSN(電子版)1611-3349

その他

その他17th International Conference on Neural Information Processing, ICONIP 2010
オーストラリア
Sydney, NSW
期間11/22/1011/25/10

Fingerprint

Action Recognition
Self organizing maps
Bins
Self-organizing Map
Descriptors
Histogram
Gradient
Local Features
Region of Interest
Human
Quantization

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

これを引用

Ji, Y., Shimada, A., & Taniguchi, R. I. (2010). Human action recognition by SOM considering the probability of spatio-temporal features. : Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings (PART 2 版, pp. 391-398). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻数 6444 LNCS, 番号 PART 2). https://doi.org/10.1007/978-3-642-17534-3_48

Human action recognition by SOM considering the probability of spatio-temporal features. / Ji, Yanli; Shimada, Atsushi; Taniguchi, Rin Ichiro.

Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2. 編 2010. p. 391-398 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻 6444 LNCS, 番号 PART 2).

研究成果: 著書/レポートタイプへの貢献会議での発言

Ji, Y, Shimada, A & Taniguchi, RI 2010, Human action recognition by SOM considering the probability of spatio-temporal features. : Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2 Edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 番号 PART 2, 巻. 6444 LNCS, pp. 391-398, 17th International Conference on Neural Information Processing, ICONIP 2010, Sydney, NSW, オーストラリア, 11/22/10. https://doi.org/10.1007/978-3-642-17534-3_48
Ji Y, Shimada A, Taniguchi RI. Human action recognition by SOM considering the probability of spatio-temporal features. : Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2 版 2010. p. 391-398. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). https://doi.org/10.1007/978-3-642-17534-3_48
Ji, Yanli ; Shimada, Atsushi ; Taniguchi, Rin Ichiro. / Human action recognition by SOM considering the probability of spatio-temporal features. Neural Information Processing: Models and Applications - 17th International Conference, ICONIP 2010, Proceedings. PART 2. 版 2010. pp. 391-398 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
@inproceedings{49cdb9c489cf434e879d688dd5ce4d9d,
title = "Human action recognition by SOM considering the probability of spatio-temporal features",
abstract = "In this paper, an action recognition system was invented by proposing a compact 3D descriptor to represent action information, and employing self-organizing map (SOM) to learn and recognize actions. Histogram Of Gradient 3D (HOG3D) performed better among currently used descriptors for action recognition. However, the calculation of the descriptor is quite complex. Furthermore, it used a vector with 960 elements to describe one interest point. Therefore, we proposed a compact descriptor, which shortened the support region of interest points, combined symmetric bins after orientation quantization. In addition, the top value bin of quantized vector was kept instead of setting threshold experimentally. Comparing with HOG3D, our descriptor used 80 bins to describe a point, which reduced much computation complexity. The compact descriptor was used to learn and recognize actions considering the probability of local features in SOM, and the results showed that our system outperformed others both on KTH and Hollywood datasets.",
author = "Yanli Ji and Atsushi Shimada and Taniguchi, {Rin Ichiro}",
year = "2010",
month = "12",
day = "21",
doi = "10.1007/978-3-642-17534-3_48",
language = "English",
isbn = "3642175333",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 2",
pages = "391--398",
booktitle = "Neural Information Processing",
edition = "PART 2",

}

TY - GEN

T1 - Human action recognition by SOM considering the probability of spatio-temporal features

AU - Ji, Yanli

AU - Shimada, Atsushi

AU - Taniguchi, Rin Ichiro

PY - 2010/12/21

Y1 - 2010/12/21

N2 - In this paper, an action recognition system was invented by proposing a compact 3D descriptor to represent action information, and employing self-organizing map (SOM) to learn and recognize actions. Histogram Of Gradient 3D (HOG3D) performed better among currently used descriptors for action recognition. However, the calculation of the descriptor is quite complex. Furthermore, it used a vector with 960 elements to describe one interest point. Therefore, we proposed a compact descriptor, which shortened the support region of interest points, combined symmetric bins after orientation quantization. In addition, the top value bin of quantized vector was kept instead of setting threshold experimentally. Comparing with HOG3D, our descriptor used 80 bins to describe a point, which reduced much computation complexity. The compact descriptor was used to learn and recognize actions considering the probability of local features in SOM, and the results showed that our system outperformed others both on KTH and Hollywood datasets.

AB - In this paper, an action recognition system was invented by proposing a compact 3D descriptor to represent action information, and employing self-organizing map (SOM) to learn and recognize actions. Histogram Of Gradient 3D (HOG3D) performed better among currently used descriptors for action recognition. However, the calculation of the descriptor is quite complex. Furthermore, it used a vector with 960 elements to describe one interest point. Therefore, we proposed a compact descriptor, which shortened the support region of interest points, combined symmetric bins after orientation quantization. In addition, the top value bin of quantized vector was kept instead of setting threshold experimentally. Comparing with HOG3D, our descriptor used 80 bins to describe a point, which reduced much computation complexity. The compact descriptor was used to learn and recognize actions considering the probability of local features in SOM, and the results showed that our system outperformed others both on KTH and Hollywood datasets.

UR - http://www.scopus.com/inward/record.url?scp=78650192992&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78650192992&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-17534-3_48

DO - 10.1007/978-3-642-17534-3_48

M3 - Conference contribution

AN - SCOPUS:78650192992

SN - 3642175333

SN - 9783642175336

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 391

EP - 398

BT - Neural Information Processing

ER -