The article presents a study that investigates the role of mood in predicting emotions, specifically in the context of human-computer interaction. The authors propose a method for time-continuous valence prediction from videos, integrating multimodal cues including mood and emotion-change labels. The study shows that both long and short-term contextual cues are crucial for emotion inference. It also underscores the importance of considering the effect of long-term affect or mood in emotion inference, a factor that has been largely overlooked in previous studies. The results confirm that using mood labels significantly improves dynamic valence prediction.
Publication date: 15 Feb 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2402.08413
