Video Temporal Grounding on Charades-STA (zero-shot transfer)
50.4R@0.5Qwen3-VL-8B + F2G-FT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen3-VL-8B + F2G-FTBackbone=Qwen3-VL-8B-Instruct, Evaluation Protocol=Foresee-to-Ground fine-tuning2026.05 | 50.4 | 71.3 | 25.7 | 47.2 | |
| Qwen3-VL-8B + FTBackbone=Qwen3-VL-8B-Instruct, Evaluation Protocol=Fine-tuning2026.05 | 44.7 | 65.8 | 21.6 | 42.9 | |
| NumProBackbone=LongVA-7B-DPO, Evaluation Protocol=Zero-shot transfer2026.05 | 42 | 63.8 | 20.6 | 41.4 | |
| TRACEBackbone=CLIP ViT-L + Mistral-7B, Evaluation Protocol=Zero-shot transfer2026.05 | 40.3 | — | 19.4 | — | |
| Qwen3-VL-8BBackbone=Qwen3-VL-8B-Instruct, Evaluation Protocol=Zero-shot transfer2026.05 | 37.7 | 65.4 | 15.9 | 40.4 | |
| VTimeLLMBackbone=CLIP ViT-L/14 + Vicuna-13B, Evaluation Protocol=Zero-shot transfer2026.05 | 34.3 | 55.3 | 14.7 | 34.6 | |
| VTG-LLMBackbone=ViT-G/14 + BLIP + LLaMA-2-7B, Evaluation Protocol=Zero-shot transfer2026.05 | 33.8 | 52 | 15.7 | — | |
| HawkEyeBackbone=-, Evaluation Protocol=Zero-shot transfer2026.05 | 31.4 | 50.6 | 14.5 | 33.7 | |
| GroundingGPTBackbone=CLIP ViT-L/14 + Vicuna-v1.5-7B, Evaluation Protocol=Zero-shot transfer2026.05 | 29.6 | — | 11.9 | — | |
| MomentorBackbone=CLIP ViT-L4 + LLaMA-7B, Evaluation Protocol=Zero-shot transfer2026.05 | 23 | 42.9 | 12.4 | 29.3 | |
| TimeChatBackbone=ViT-G/14 + LLaMA-2-7B, Evaluation Protocol=Zero-shot transfer2026.05 | 22.9 | 47.7 | 12.5 | 30.6 |