Automatic video surveillance system: recognition of critical behaviors in the process of machine learning

This article describes means for interpreting and parametrization of critical behaviours. This is the basis for developing the database necessary for the implementation of an automated video surveillance system. The database will be used in the machine learning process as a decision block for the entire system. To this end, it is necessary to conduct an in-depth analysis of events that indicate a potential threat to human health or life. The project requires defining a hardware and software platform and how to integrate them into an optimal system. For the detection and tracking of human skeletons, it was decided to use the popular Microsoft Kinect 2.0 depth sensor, used in a wide spectrum of recreational games and applications. Integration with the platform, based on an artificial neural network, was decided to be developed in the well-known MATLAB programming environment. The author reveals details on the process of analysing potential emergency situations, parameterization, performing empirical measurements and developing a database.