A Behavior Recognition Method for Elderly People Based on Background Cumulative Subtraction and YOLOv5 Algorithm
Xuan Yin, Bin Zhao, Linshuo Wu, Shaokang Wang, Jun Feng, Binyu Xiong, Yijia Chen · 2024
To solve the problem of elderly people who are living alone and unable to receive timely assistance in the event of accidents and are prone to death, this paper proposed a behavior recognition method based on background cumulative subtraction method and YOLOv5 algorithm. Based on the filmed videos, the background cumulative subtraction method is proposed for motion object detection. The experimental results show that this method can effectively improve the detection of IoU and improve accuracy. Then, a behavior recognition model is trained by YOLOv5 algorithm. The model achieved a mAP of 0.983 at an IoU threshold of 0.5, and an F1Score of 0.96 at a confidence threshold of 0.443, indicating high model accuracy. Finally, this paper combines the background cumulative difference method and YOLOv5 algorithm to improve the speed of behavior recognition. The experimental results show that the proposed algorithm can reduce the detection time of each frame by 8.47% and improve real-time performance.