AI Paper Atlas

Research map / Video / Video-language understanding

EventSTU: Event-Guided Efficient Spatio-Temporal Understanding for Video-LLMs

Wenhao Xu, Xin Dong, Yue Li, Haoyuan Shi, Yueyi Zhang, Zhiwei Xiong

ECCV 2026 · Video › Video-language understanding · approach: Long video understanding · 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
add adaptive retention ratio to ZAP (ablation)ZAP · EventBench / LongVideoBenchaccuracy: 54.09 -> 55.09 (EventBench); 59.54 -> 61.11 (LongVideoBench)Improved
combine physics-aware and semantic-aware selection (ablation)ZAP · EventBenchaccuracy: 52.10 -> 54.09Improved
add physics-aware selection to semantic-only (ablation)ZAP · LongVideoBenchaccuracy: 58.12 -> 59.54Improved
accumulation (ACC) coarse sampling instead of uniform (ablation)C2FS · EventBenchaccuracy: 53.09 -> 55.09Improved
accumulation (ACC) coarse sampling instead of top density (ablation)C2FS · EventBenchaccuracy: 54.09 -> 55.09Improved
bin sampling (BIN) instead of TOP at fine stage (ablation)C2FS · EventBenchaccuracy: 53.89 -> 55.09Improved
add C2FS keyframe sampling instead of AKSLLaVA-OV-7B · Avg (4 benchmarks)average score: 61.06 -> 61.59Improved
add ZAP token pruning instead of DyCoke at 50%LLaVA-OV-7B · Avg (4 benchmarks)average score: 57.97 -> 59.07Improved
add EventSTU at 30% retentionLLaVA-OV-7B · Avg (4 benchmarks)average score: 58.05 -> 61.86Improved
add EventSTU at 30% retentionQwen3-VL-8B · Avg (4 benchmarks)average score: 66.60 -> 69.29Improved
raise coarse sampling ratio S to 0.50 (ablation)C2FS · number-of-frames sweepaccuracyHurt
add ZAP at high retention ratio R=0.7 (ablation)LLaVA-OV-7B · Fig 10accuracy: +2.19 pointsImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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