An effective model for intergrated face detection and recognition

Abstract

Facial recognition, an attractive field in computer-based application, has been one of the most widely research and challenging areas in computer vision and machine learning. The innovation of new face authentication technologies is a controversal topic to build much effective and robust face recognition algorithms. In this work, an effective, fast and reliable model is proposed based on combining traditional algorithms such as HOG, SVM and the modern ones such as ResNet50, Facial Landmark 68 for face recognition and emotion detection. Tests on different databases of large number of samples, various environmental conditions and facial expressions are presented with high recognition results.