MMSI

The 'MMSI' dataset/benchmark contains multimodal data for evaluating social interactions in video, focusing on the alignment of verbal and non-verbal cues among multiple speakers.

Papers using MMSI (3)

When and How Much to Imagine: Adaptive Test-Time Scaling with World Models for Visual Spatial Reasoning2026Multi-speaker Attention Alignment for Multimodal Social Interaction2025Scaling Spatial Intelligence With Multimodal Foundation Models2025