An Effective Emotion Recognition Method Using Facial and Speech Features

In face-to-face conversations, facial expressions help humans communicate nonverbally. Since the early 1990s, academics' focus on automatic facial expression detection, which is essential for human-machine interfaces, has grown. Traditional machine learning techniques can demand a difficult feature extraction procedure and yield subpar results. Using current developments in deep learning, we present efficient deep Convolutional Neural Networks (CNNs) that can automatically and accurately evaluate the semantic information contained in faces and speech without manually defining feature descriptors. Additionally, in order to develop an effective CNNs based emotion recognition system with a high degree of classification capability, we employ various loss functions and training techniques. The experimental results demonstrate that, the proposed method outperforms existing solutions in terms of performance and makes it ideal for real-time systems.