Dual Phase Learning for Large Scale Video Gait Recognition

• Main Authors: SHEN, Jialie, PANG, Hwee Hwa, TAO, Dacheng, LI, Xuelong
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2010
• Subjects: Databases and Information Systems; Numerical Analysis and Scientific Computing
• Online Access: https://ink.library.smu.edu.sg/sis_research/503; https://ink.library.smu.edu.sg/context/sis_research/article/1502/viewcontent/Dual_Phase_Learning_for_Large_Scale_Video_Gait_Recognition__edited_.pdf
• Institution: Singapore Management University

Accurate gait recognition from video is a complex process involving heterogenous features, and is still being developed actively. This article introduces a novel framework, called GC2F, for effective and efficient gait recognition and classification. Adopting a ”refinement-and-classification” principle, the framework comprises two components: 1) a classifier to generate advanced probabilistic features from low level gait parameters; and 2) a hidden classifier layer (based on multilayer perceptron neural network) to model the statistical properties of different subject classes. To validate our framework, we have conducted comprehensive experiments with a large test collection, and observed significant improvements in identification accuracy relative to other state-of-the-art approaches.