Scalable Emotion Recognition Model with Context Information for Driver Monitoring System

Understanding emotions from an individual's perspective is critical for daily social interactions. If machines could similarly comprehend emotions, they could interact more effectively with people. Recognizing emotions accurately often necessitates considering the situational context, which helps in identifying a broader spectrum of emotions. Current emotion detection systems predominantly rely on facial images, often overlooking contextual influences. This paper proposes an emotion recognition model that combines facial feature analysis with an understanding of the surrounding context. The validation on the EMOTIC benchmark confirms the model's usefulness, registering an overall accuracy percentage of 84.9%. The paper emphasizes the necessity of combining contextual information for more accurate emotion recognition, which will pave the way for advances in sectors such as medical imaging, augmented reality, and human-computer interaction.