Face detection and recognition in video surveillance systems

Abstract

The article discusses modern approaches to research the tasks of face detection and recognition in biometric video surveillance systems. Intelligent systems based on biometrics are becoming more widespread every year in various spheres of human life – from security systems to banks and shops. Various unique biological identifiers of a person can act as biometric data - fingerprints, retinal pattern, skin texture, handwriting, and more. But the most widespread systems are based on facial recognition. Such systems are characterized by minimal hardware requirements: it is enough to place a video surveillance camera and the ease of implementation of the recognition algorithm. The most common algorithms for detecting and recognizing faces and the requirements for data sets used for training models will be considered in this paper. The first section describes the concept of a face as an identifier in a biometric recognition system. The second section describes different data sets used to train models for face detection and recognition. The third section contains a description of the basic common models and libraries of face detection and recognition. The final section provides an example of the structure of a biometric facial recognition system.