FACIAL RECOGNITION SYSTEM USING HETEROGENEOUS PROGRAMMING

Face recognition technology is a biometric technology that uses a person’s face traits to identify them. Face photos are collected by people, and the recognition technology processes them automatically. It starts with detection, which involves differentiating human faces from other things in the picture, and then moves on to recognizing those faces that have been discovered. In this research project, a deep learning model based on the ResNet Network is created to recognize the existence of a concerned individual’s face kept in the database. The most common HOG Classifier was used to complete the detection task, as it is lighter and more accurate than other detectors currently in use. Face detection and identification is a difficult process that needs a lot of computational resources .We experimented with highly-parallel Graphics Processing Units. We have also compared the results of a typical CPU based implementation with our GPU based implementation to chalk out the efficiency and factor of speed-up. We have also developed a Graphical Interface using the Qt Platform to provide ease of usage.