Depression mood gait recognition method and system based on double-branch self-erasure graph convolutional network

• Main Authors: QIU SONG, JIAO YUEJUN, FENG RUI, ZHANG BOYUAN, HE XIAOWEN, XU YUAN, MA WEIJUN
• Format: Patent
• Language: Chinese; English
• Published: 09.01.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses a depression mood gait recognition method and system based on a double-branch self-erasure graph convolutional network. The system comprises a data acquisition module, a data preprocessing module, a model construction training module and a detection result output module. According to the method, psychological test data labels of a subject and walking gait data are collected to construct a data set, and features are extracted through a network, so that a depressive emotion detection model is obtained. An original gait video is processed into a scaling gait period video, a 3D-ResNet network is used for extracting shallow layer features, DSAM-GN is used for extracting deep layer features, and finally a result is obtained through a classifier. The method is high in precision and few in model parameters, and human body gait features can be better extracted; the method is high in real-time performance, low in cost and suitable for screening the mental health conditions of people, and effecti