Image Recognition of Prohibited Express Packages Based on YOLOv4

In the increasingly complex security environment of the express industry, the identification problem of prohibited items with small samples and multiple labels has not been properly solved. These prohibited items greatly increase the risk of safe transportation of express delivery and bring great security risks to the society. The key to solve this problem is to realize accurate and efficient identification of prohibited items with small samples and multiple labels. On this basis, the YOLOv4 algorithm model was used to realize the recognition of security check images of the delivered items, and the loss function value of the best model was trained to be 0.95 and the accuracy was 75%.