VISTA (Video Grounding) - Human-Human: leaderboard
Metric: Mean spatio-temporal IoU (m_vIoU, %): per-frame box IoU summed over the frames where predicted and true tubes overlap in time, divided by their temporal union, on human-human interactions, over the 11,814 video-query pairs of VISTA (HCSTVG-v1 and v2, VidVRD, VidSTG, MeViS and RVOS videos re-annotated with an interaction taxonomy), zero-shot on sub-sampled frames; the model must localize the queried subject with a box in every frame; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen3-VL-8B | 66.2 |
| 2 | InternVL2.5-8B | 49.2 |
| 3 | Qwen-VL-Chat | 48.2 |
| 4 | VISTA Sphinx-v2 (checkpoint unspecified) | 47 |
| 5 | MiniGPT-v2 | 46 |
| 6 | VISTA MiMo-VL-7B (checkpoint unspecified) | 45.1 |
| 7 | VISTA CogVLM Grounding (checkpoint unspecified) | 44.7 |
| 8 | Shikra-7B | 34 |
| 9 | LLaVA-Grounding-7B | 28.4 |
| 10 | Ferret-7B | 26.4 |
Interactive version: theaggregate.ai/benchmark?slug=vista-video-grounding-human-human · How It Works · Data refreshed daily, snapshot 2026-10-07.