Face Detection and Recognition, Face Emotion Recognition Through NVIDIA Jetson Nano

摘要

This paper focuses on implementing face detection, face recognition and face emotion recognition through NVIDIA's state-of-the-art Jetson Nano. Face detection is implemented using OpenCV's deep learning-based DNN face detector, supported by a ResNet architecture, for achieving better accuracy than the previously developed models. The result computed by framework libraries of OpenCV, with the support of the above-mentioned hardware, displayed reliable accuracy even with the change in lighting and angle. For face recognition, the approach of deep metric learning using OpenCV, supported by a ResNet-34 architecture, is used. Face emotion recognition is achieved by developing a system in which the areas of eyes and mouth are used to convey the analysis of the information into a merged new image, classifying the image into displaying any of the seven basic facial emotions. A powerful and a low-power platform, Jetson Nano carried out intensive computations of algorithms easily, contributing in high video processing frame.