Action Recognition Based on Weight BOVW

DOI

10.2991/ncce-18.2018.56How to use a DOI?

Keywords

bag of words; action recognition; SVM.

Abstract

To improve the accuracy and robustness of human action recognition, a kind of human action recognition algorithm based on the weighing bad-of-word model was put forward. The features extracted from video sequences through existing algorithms contained HOG feature, HOF feature, and MBH feature, and relevant visual vocabulary table was acquired through the K-means clustering method. Statistics of the word frequency of all visual words of each category was made, and the words that rank at the front in the word frequency were selected to conduct the weighted normalization processing and acquire the weight of the words. The test samples conducted weighing expressions on the words whose frequency ranked at the front positions during the generalization process of bad of words, and it conducted classification and recognition of the characteristics of the bag of words through SVM. The result of UCF Sports DS experiment showed that the new algorithm had a good recognition capacity

Copyright

© 2018, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Taizhe Tan
AU  - Chuhong Li
PY  - 2018/05
DA  - 2018/05
TI  - Action Recognition Based on Weight BOVW
BT  - Proceedings of the 2018 International Conference on Network, Communication, Computer Engineering (NCCE 2018)
PB  - Atlantis Press
SP  - 357
EP  - 361
SN  - 1951-6851
UR  - https://doi.org/10.2991/ncce-18.2018.56
DO  - 10.2991/ncce-18.2018.56
ID  - Tan2018/05
ER  -