Learning and Classification of Events in Monitored Environments

This paper presents a prototype system to automatically carry out surveillance tasks in monitored environments. This system consists in a supervised machine learning algorithm that generates a set of highly interpretable rules in order to classify events as normal or anomalous from 2D images without needing to build a 3D model of the environment. Each security camera has an associated knowledge base which is updated when the environmental conditions change. To deal with uncertainty and vagueness inherent in video surveillance, we make use of Fuzzy Logic. The process of building the knowledge base and how to apply the generated sets of fuzzy rules is described in depth for a virtual environment. Keywords— Automated Video Surveillance, Visual Information Analysis, Machine Learning, and Fuzzy Logic.