Biometrické softwarové prostředí pro identifikaci, klasifikaci a sledování uživatele na základě analýzy obrazu obličeje

Abstract

This diploma thesis deals with the design and realization of a hybrid biometric system for face detection, tracking and recognition from static and dynamic images. The thesis describes different methods of face detection, tracking and recognition from 2D data. The thesis involves the process of realizing the biometric system using Viola-Jones, KLT algorithm and pretrained AlexNet convolutional neural network. In addition, the thesis includes also objective testing of created system towards the variable degradation effects and describes the realized SW environment for user classification purposes.