COSMI: 222K Multi-Object Interaction Sequences via Compositional Synthesis

University of Tübingen's Gerard Pons-Moll team released COSMI, a dataset of 222K multi-object human-object interaction sequences built by composing local single-object interactions, used to train a diffusion Transformer.

2026-10-07 ~ 2026-10-08 · 2 related posts

• COSMI composes single-object captures into 222k multi-object interaction sequences, 30x larger than prior sets — UniTuebingen · 2026-10-07
• COSMI composes single-object HOI datasets into 222k multi-object sequences to train a diffusion transformer — CSProfKGD · 2026-10-08