Researchers have developed SkillMoV, a novel framework for estimating human proficiency from multi-view video. This parameter-efficient system utilizes a Mixture-of-View Projector (MoVP) that adapts the mixture-of-experts paradigm to different camera viewpoints. SkillMoV incorporates soft routing, cross-view attention, prototype anchoring, and gated projection to produce skill embeddings. When evaluated on the EgoExo4D dataset, SkillMoV achieved a 50.17% overall accuracy in the Exos setting, outperforming existing methods by 3.57 percentage points with a single, jointly trained model. AI
IMPACT Enhances automated skill assessment in diverse fields by improving multi-view video analysis.
RANK_REASON The cluster contains a research paper detailing a new framework and its evaluation on a dataset.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
