A Fall Detection Alert System Based on Lightweight Openpose and Spatial-Temporal Graph Convolution Network
Weijian Mai, Fengjie Wu, Ziqian Guo, Yuhan Xiang, Gensheng Liu, Xiaobin Chen · Journal of Physics Conference Series · 2021
To solve the home safety problem of the solitary group, we have developed a system based on Lightweight Openpose and Spatial-Temporal Graph Convolution Network (ST-GCN). We use Lightweight Openpose to extract the human skeleton sequence from the video stream in real-time, and use ST-GCN to extract the action time and space features from it to obtain an efficient fall detection model. For further application, we develop an email alert system using STMP protocol. This work integrates fall detection into the surveillance camera to realize real-time intelligent monitoring and detection at home. When the family member falls, the picture will be captured, and the alarm email will be sent to the owner or guardian, to provide a safer guarantee for the family member, especially the empty-nesters and young people living alone.