AI Paper Atlas

Research map / Video / Video-language understanding

Foresee-to-Ground: From Predictive Temporal Perception to Evidence-Driven Reasoning for Video Temporal Grounding

Zelin Zheng, Xinyan Liu, Ruixin Li, Antoni B. Chan, Guorong Li, Qingming Huang, Laiyun Qing

ICML 2026 · Video › Video-language understanding · approach: Temporal video grounding · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace conventional fine-tuning with F2G fine-tuningQwen3-VL-8B · Charades-STA (zero-shot)R@0.7: 21.6 -> 25.7Improved
replace conventional fine-tuning with F2G fine-tuningQwen3-VL-8B · ActivityNet-CaptionsmIoU: 40.8 -> 45.7Improved
replace conventional fine-tuning with F2G fine-tuningQwen3-VL-8B · QVHighlights (zero-shot)HIT@1: 36.8 -> 45.6Improved
apply F2G-FT to other Video-LLM backbonesLLaVA-NeXT-7B · ActivityNet-CaptionsmIoU: 12.6 -> 23.3Improved
apply F2G-FT to other Video-LLM backbonesQwen2.5-VL-7B · Charades-STAR@0.7: 10.3 -> 14.2Improved
add Stage-1 predictive pretraining of the temporal module (ablation)F2G · ActivityNet-CaptionsmIoU: 41.8 -> 45.7Improved
add SIGReg regularizer in Stage-1 (ablation)F2G · ActivityNet-CaptionsR@0.7: 26.8 -> 28.4Improved
add ID citation (Identify step) before timestamps (ablation)F2G · Charades-STA (zero-shot)R@0.7: 21.5 -> 25.7Improved
add segment visual evidence tokens Pk to evidence units (ablation)F2G · ActivityNet-CaptionsR@0.7: 23.5 -> 28.4Improved
increase Top-K from 4 to 8 (ablation)F2G · ActivityNet-CaptionsR@0.7: 25.1 -> 28.4Improved
increase SEE queries from 4 to 8 (ablation)F2G · ActivityNet-CaptionsmIoU: 45.7 -> 45.2No clear effect
fine-tune with F2G on VTG dataQwen3-VL-8B · VideoMME (w/o subtitles)overall accuracy: 71.4 -> 70.9No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy