HUG-VIS - Emotion Recognition (Video): leaderboard

Metric: Seven-class emotion recognition accuracy (%; 8,400 seated half-body clips of 30 professional actors, each performing the same 280 emotion-action-prompt assignments in a controlled Mandarin studio, 1,200 clips per emotion; zero-shot; input: video). Source: arxiv.org. Saturation forecast: Around December 2026. 7 models tracked.

Top models

#ModelScore
1MiniCPM-o-4.548.52
2Molmo2-8B46.34
3Qwen2.5-Omni-7B27.77

Interactive version: theaggregate.ai/benchmark?slug=hug-vis-emotion-recognition-video · How It Works · Data refreshed daily, snapshot 2026-09-29.