Image Infringement Judgement with CNN-based Face Recognition

Image Infringement Judgment can help public figures to solve infringement of portrait rights. This paper uses Convolutional Neural Network (CNN) to predict the category of images. First, we use data scraping to get public photos to create the data set required for the training model. Secondly, this paper constructs a face prediction model through the CNN framework and extracts the feature of faces through the algorithm. Finally, we adjust parameters to optimize the model through data visualization. To verify the effectiveness of CNN, we compare it with the Multilayer Perceptron (MLP) on two different data sets, one is the data set built by reading camera pictures, and the other is the data set built by static pictures. The experimental results show that the face recognition application based on the CNN model proposed in this paper has an accuracy rate of 97%. It can classify the extracted data set well, can obtain the data content, and make accurate predictions. We applied this accurate model to the judgment of the infringement of images on the WeChat public account.