FPGA implementation of emotion recognition

Abstract

Many children with autism are faced with the difficulty in understanding how people feel. This impairs communication between them and others resulting in drawback from communication. Therefore, a portable device that can detect facial emotion was suggested to help the autistic children. This project aims to create an emotion detector using FPGA. The emotion detector will display the emotion of the person based on the image given. Sobel filter and skin detection were used to crop the image of the mouth before calculating the angle to determine the emotion. The 2 most important factors that are needed for the angle calculation are the edges of the mouth and the middle top of the mouth. Although the program is only able to detect happy and sad emotions, up to 85% of precision was achieved for the mouth detection by the program.