Cross- View Gait Recognition Using Non-Linear View Transformations of Spatiotemporal Features

摘要: This paper presents a novel cross-view gait recognition technique based on the spatiotemporal characteristics of human motion. We propose deep fully-connected neural network with unsupervised learning which transfers descriptors from multiple views to single canonical view. The proposed non-linear learns model for all videos captured different viewpoints and finds shared high-level virtual path map them Therefore, does not require any labels or viewpoint information in phase. is learned only once using motion features sequences several viewpoints, later it used construct gallery probe sets. are classified simple linear support vector machine. Experiments carried out benchmark dataset, CASIA-B, comparisons state-of-the-art demonstrate that method outperforms existing algorithms.