Indoor Falling Behavior Detection Algorithm Based on Improved YOLOv7
To address the problem of detecting falls for the elderly people in indoor surveillance video,a real-time fall behavior de-tection algorithm based on improved YOLOv7 network model was proposed.the strided convolution is traditionally used in the target detection model based on YOLOv7 to realize the downsampling feature,but this perhaps make the feature of the target information fuzzy.To solve this problem,a novel downsampling module,robust feature downsampling,is introduced to improve the clarity of target information features during downsampling.In addition,by introducing the Coord Attention attention mechanism in the concat section of the network,the spliced feature graphs can be better merged.Experimental results show that the improved YOLOv7 model has a high accuracy and detection performance in falling behavior detection,with an accuracy of 98.88％,mAP50 value of 98.83％,and mAP50∶95 value of 74.12％.This means that the algorithm can accurately detect the fall behavior of the elderly,so the family should promptly discover and make necessary rescue measures in a timely manner.
万方数据
维普
