Human Action Recognition Based on Multi-view Semi-supervised Learning
TANG Chao1, WANG Wenjian2, WANG Xiaofeng1, ZHANG Chen1, ZOU Le1
1.Department of Computer Science and Technology, Hefei University, Hefei 230601 2.School of Computer and Information Science, Shanxi University, Taiyuan 030006
Abstract Since human action is complicated in nature, single action feature view lacks the ability of comprehensively profiling human action. A method for human action recognition based on multi-view semi-supervised learning is proposed in this paper. Firstly, a method based on three different modal views is proposed to represent human action, namely Fourier descriptor feature view based on RGB modal data, spatial and temporal interest point feature view based on depth modal data and joints projection distribution feature view based on joints modal data. Secondly, multi-view semi-supervised learning framework is utilized for modeling. The complementary information provided by different views is utilized to ensure better classification accuracy with a small amount of labeled data and a large amount of unlabeled data. The classifier-level fusion technology is employed to combine the predictive ability of three views, and the problem of confidence evaluation of unlabeled samples is effectively solved. The
Fund:Supported by National Natural Science Foundation of China(No.61673249,61672204,61806068), Natural Science Foundation of Anhui Province(No.1908085MF184), Key Scientific Research Foundation of Education Department of Anhui Province(No.KJ2018A0556,KJ2018A0555), Excellent Talents Training Funded Project of Universities of Anhui Province(No.gxfx2017099,gxyqZD2017076), Scientific Research Fund Project of Talents of Hefei University(No.16-17RC19), Key Teaching and Research Project of Hefei University(No.2018 hfjyxm09)
About author:: TANG Chao(Corresponding author), Ph.D., associate professor. His research interests include machine learning and compu-ter vision.WANG Wenjian, Ph.D., professor. Her research interests include machine learning and computing intelligence.WANG Xiaofeng, Ph.D., professor. His research interests include image processing and computing intelligence.ZHANG Chen, Ph.D., lecturer. Her research interests include machine learning and computing intelligence.ZOU Le, Ph.D. candidate, associate professor. His research interests include image processing and computing intelligence.
[1] LÜ M Q, CHEN L, CHEN T M, et al. Bi-view Semi-supervised Learning Based Semantic Human Activity Recognition Using Acce-lerometers. IEEE Transactions on Mobile Computing, 2018, 17(9): 1991-2001. [2] YUAN H J. A Semi-supervised Human Action Recognition Algorithm Based on Skeleton Feature. Journal of Information Hiding and Multimedia Signal Processing, 2015, 6(1): 175-182. [3] LIU W F, LIU H L, TAO D P, et al. Multiview Hessian Regula-rized Logistic Regression for Action Recognition. Signal Processing, 2015, 110: 101-107. [4] WANG S, MA Z G, YANG Y, et al. Semi-supervised Multiple Feature Analysis for Action Recognition. IEEE Transactions on Multimedia, 2014, 16(2): 289-298. [5] ZHANG T Z, LIU S, XU C S, et al. Boosted Multi-class Semi-supervised Learning for Human Action Recognition. Pattern Recogni-tion, 2011, 44(10/11): 2334-2342. [6] LAPTEV I, LINDEBERG T. Space-Time Interest Points // Proc of the 9th IEEE International Conference on Computer Vision. Wa-shington, USA: IEEE, 2003: 432-439. [7] HARRIS C, STEPHENS M. A Combined Corner and Edge Detector // Proc of the 4th Alvey Vision Conference. New York, USA: ACM, 1988, 15: 147-151. [8] DOLLAR P, RABAUD V, COTTRELL G, et al. Behavior Recognition via Sparse Spatio-Temporal Features // Proc of the 2nd Joint IEEE International Workshop on Visual Surveillance and Perfor-mance Evaluation of Tracking and Surveillance. Washington, USA: IEEE, 2005: 65-72.
2019, Vol. 32 
