VISTA (Video Grounding) - Human-Object: 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-object 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 | 64.7 |
| 2 | VISTA CogVLM Grounding (checkpoint unspecified) | 54 |
| 3 | InternVL2.5-8B | 49.5 |
| 4 | Qwen-VL-Chat | 47.9 |
| 5 | VISTA Sphinx-v2 (checkpoint unspecified) | 46.8 |
| 6 | MiniGPT-v2 | 46.1 |
| 7 | LLaVA-Grounding-7B | 37 |
| 8 | VISTA MiMo-VL-7B (checkpoint unspecified) | 36.1 |
| 9 | Shikra-7B | 35.3 |
| 10 | Ferret-7B | 24.5 |
Interactive version: theaggregate.ai/benchmark?slug=vista-video-grounding-human-object · How It Works · Data refreshed daily, snapshot 2026-10-07.