Abstract
This paper will discuss current approaches to the recognition of violent human actions. Recognition accuracy for these methods will also be presented. In the modern world due to development of sensory and visual technologies systems of recognition of violent actions of the person have become popular. The problem of aggression, including children's aggression, has long attracted the attention of psychologists. Among methods using violence there are several separate methods: methods of detection of violence using machine learning, methods of detection of violence using SVM and methods of detection of violence using deep learning. The purpose of this study is to review current software solutions for recognizing violent movements from video. During the work done, an analysis of existing machine learning and artificial intelligence algorithms for automatic detection of aggression was made. And also the application of the PoseNet method for aggression detection was tested and proposed. An attractive feature of the algorithm is that its performance is not affected by the number of people in the input image. Pose evaluation models take the processed camera image as input data and output information about the key points.