Human body falling detection method, device and system

• Main Authors: ZHAI DENGHUI, LI ZHAOYANG, XI PIQI, SHI HONGJUN, ZHANG LEI, ZHENG WEIHUA, ZHANG YANLONG, LI MINGDAO, ZHAO PANDIAN, HE XINGDUO, WANG XING
• Format: Patent
• Language: Chinese; English
• Published: 28.05.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; PHYSICS

The invention belongs to the technical field of human behavior detection, and particularly relates to a human falling detection method, device and system. The method comprises the following steps: firstly, acquiring a field operation video and performing framing processing; secondly, extracting feature points of the staff from each frame of the obtained field operation image by using an AlphaPose model, and obtaining a skeleton image of the staff in each frame of the field operation image according to a feature point extraction result; dividing all skeleton images into a plurality of segments; extracting a frame of skeleton image and an optical flow image between a frame of skeleton image and an adjacent frame of skeleton image from each section of skeleton image; and finally, respectively inputting a frame of skeleton image and a frame of optical flow image extracted from a certain section of skeleton image into a space flow channel and a time flow channel in the double-flow network model, thereby obtaining