Falling Behavior Detection Based on Improved YOLOv5s
In order to achieve the real-time detection of fall behavior among power plant personnel,and prevent them from falling and falling into a coma,it cannot be detected and rescued in a timely manner,an improved YOLOv5s falling behavior detection algo-rithm network is proposed to address the issues of insufficient real-time detection and feature extraction capabilities.The introduction of the SKAttention module in the YOLOv5s model enables the network to automatically utilize the information captured by the effec-tive receptive fields for classification.This new deep structure allows the CNN to perform dynamic selection mechanisms on the convo-lutional core,thereby adaptively adjusting the size of its receptive field;By combining adaptively spatial feature fusion(ASFF)and fully utilizing different features,and introducing weight parameters into the algorithm,based on multi-level functions,the accuracy of underwater target recognition is achieved;The spatial pyramid pooling structure with context sensitive pyramid convolution(SPPFC-SPC)is added to greatly reduce inference time.Experimental results show that compared to the original YOLOv5s,the new network improves the mean average precision(mAP)by 2.1％,and the recall rate by 16％.The improved network is more powerful in the perception of details and spatial modeling,and can more accurately capture the abnormal falling behaviors,significantly improving a detection effect.
万方数据
维普
