Sequential convolutional network for behavioral pattern extraction in gait recognition

As a unique and promising biometric, video-based gait recognition has broad applications. The key step of this methodology is to learn the walking pattern individuals, which, however, often suffers challenges extract behavioral feature from sequence directly. Most existing methods just focus on either appearance or motion pattern. To overcome these limitations, we propose sequential convolutional network (SCN) novel perspective, where spatiotemporal features can be learned by basic backbone. In SCN, information extractors (BIE) are constructed comprehend intermediate maps in time series through templates relationship between frames analyzed, thereby distilling Furthermore, multi-frame aggregator SCN performs integration whose length uncertain, via mobile 3D layer. demonstrate effectiveness, experiments have been conducted two popular public benchmarks, CASIA-B OU-MVLP, our approach demonstrated superior performance, comparing with state-of-art methods.