A Comprehensive Study of Facial Expression Recognition

Abstract

Facial expression recognition (FER) is a key area of research interest in the field of human-computer interaction. It can make an intelligent system feel and respond to humans’ emotions. First, this review introduces FER’s important significance in diverse domains such as healthcare, smart environments, and humanrobot interaction, where emotion-aware systems can enhance communication efficiency. The review analyzes key technical components of FER systems, including dataset acquisition and preprocessing, feature extraction mechanisms, and classification models. It also discusses transfer learning, dropout regularization, and optimization strategies to improve model performance and reduce overfitting. In addition, it evaluates common performance metrics in FER research and points out the model’s limitations and advantages in different emotions, such as anger, happiness, sadness, surprise, fear, and hate. It proposes data augmentation and advanced network designs to solve problems in FER, including variations in lighting, pose, and cultural divide. Finally, summarizes and points out research interests in the future, such as the fusion of multimodal data and the development of lightweight models for real-time applications. This review can promote the use of FER in daily life. Provides reference to researchers and workers in FER, and helps to explore innovation to push the development of emotional identity tech.