Research on Dangerous Area Personnel Detection Algorithm Based on 5G Cloud Edge Collaboration

To strengthen the safety protection of personnel in deep water dangerous areas such as reservoirs and rivers, Relying on the technical characteristics of 5G network technology, such as high speed, low delay, high bandwidth, and high flexibility, Build a video surveillance network architecture; The cloud-edge collaboration technology based on 5G network sinks the cloud intelligent analysis and computing power to the edge side, The edge end realizes the comprehensive perception and real-time analysis of video data, Accelerate the analysis of real-time data, Form a scientific and effective decision-making and intelligent analysis; The cloud is responsible for storing the high-value data uploaded on the edge side, And applied the data to the training of the deep-learning models, Finally, the trained model is distributed to the edge terminal; In this paper, by replacing the SiLU activation function of the YOLO V5s original model as the Mish activation function, Improve the detection and identification rate of remote personnel in the video surveillance area; Improving the weighting factor for non-maximum suppression is for the CIoU function, Further improve the positioning and identification of personnel in the image, Finally, realize the detection and identification of personnel in dangerous areas and remote alarm, Avoid safety accidents.