Facial Expressions Recognition System Using FPGA-Based Convolutional Neural Network

Emotion detection has become one of the most significant aspects to consider in any project related to Affective Computing. There are several researches in emotion recognition,where Convolutional Neural Network is considered as an efficient algorithm that achieves the state-of-the-art performance in image recognition. There have been plenty of methods applying CNN model and conducting in software. However, traditional software based computation is not fast enough to meet the demand of real-time image processing. Therefore, we propose an efficient method for expressions recognition using FPGA-based. This model is used for classifying seven elementary types of human emotions: angry, fear, disgust, happy, sad and neutral by extracting features through those layers respectively. We also tested the functions of this model on FPGA simulation and achieved over 65% of accuracy using FER2013 dataset.