Research on Detection Method of Daily Staff Work Management Violation based on Convolutional Neural Network and SSD Algorithm

Due to high costs and low coverage, manual inspectors cannot meet the needs of normalized daily staff work management. This paper proposes a detection method for daily staff work management violations based on human behavior recognition in the video to deal with this problem. Based on the convolutional neural network and SSD algorithm, the method detects the clothing of personnel, the use of items, and the possession of police equipment in the video to achieve the purpose of daily staff work management violations. The core of the research uses the VGG16 network as the basic network structure of the SSD algorithm and conducts training on violation samples. The loss function is further replaced by focal loss, and a better network model is obtained to improve the detection performance. This paper provides the technical basis for establishing a scalable technical framework for violation detection and technical support for the intelligence and normalization of daily staff work management.