Vibe360: Group-Based Deep Emotional Understanding of Immersive Communications

Immersive communication systems enable remote attendees to experience events as if physically present. This paper examines group-based emotion recognition from 360-degree video as a potential enhancement for these systems. In a preliminary study, we recorded a 360-degree video with group-level emotion annotations and analyzed it using two advanced deep learning models for face and emotion recognition, providing both categorical and valence-arousal metrics. Results indicate that this approach effectively identifies group-level emotional shifts (e.g., from neutral to positive or negative), though finer distinctions may require additional data sources. This tool could improve immersive communication by supporting audience en-gagement monitoring and enhancing accessibility for individuals with challenges in emotion perception.