Dangerous Wasp Image Recognition Process based on Residual Neural Network

This paper combines image processing and artificial intelligence to realize the image recognition of dangerous wasps. By combining SVM, KNN and Decision Tree, we establish the Integrated Algorithm Model. Then we establish the Long Tail Residual Neural Network, which analyzes the images from the public. This network makes the margin between sparse classes as large as possible by using the new loss function so that the sparse classes have a greater weight, which has a better training effect on the unbalanced class. After the picture is inputted, the model will output the possibility that the hornets in the picture belong to each label, making the result more accurate. This research can help quickly identify dangerous wasps and protect people's lives.