AI Paper Atlas

Research map / Video / Video-language understanding

Keyframe-oriented Vision Token Pruning: Enhancing Efficiency of Large Vision Language Models on Long-Form Video Processing

Yudong Liu, Jingwei Sun, Yueqian Lin, Jianyi Zhang, Jingyang Zhang, Ming Yin, Qinsi Wang, Hai Li, Yiran Chen

ICCV 2025 · Video › Video-language understanding · approach: Long video understanding · 9 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 KVTP keyframe-oriented pruning ratesPruMerge (7B) · VideoMMEaccuracy: 59.77 -> 63.29Improved
Add KVTP keyframe-oriented pruning ratesPruMerge (7B) · EgoSchemaaccuracy: 52.49 -> 54.71Improved
Add KVTP keyframe-oriented pruning ratesPruMerge (7B) · NextQAaccuracy: 75.58 -> 76.76Improved
Add KVTP keyframe-oriented pruning ratesToMe (7B) · VideoMMEaccuracy: 58.90 -> 62.36Improved
Add KVTP keyframe-oriented pruning ratesRandom Sampling (7B) · VideoMMEaccuracy: 58.28 -> 60.16Improved
Add KVTP keyframe-oriented pruning ratesPruMerge (72B) · VideoMMEaccuracy: 64.52 -> 67.34Improved
Add KVTP keyframe-oriented pruning ratesToMe (72B) · NextQAaccuracy: 76.45 -> 79.51Improved
Add KVTP keyframe-oriented pruning ratesRandom Sampling (72B) · EgoSchemaaccuracy: 60.47 -> 62.19Improved
Use KVTP+PruMerge instead of FastV on 7B (FLOPs x36% vs x64%)LLaVA-Video-7B · NextQAaccuracy: 77.89 -> 76.76Hurt
Use soft frame selection (KVTP) instead of hard selection (KeyVideoLLM)LLaVA-Video-7B · VideoMMEaccuracy: 51.32 -> 63.29Improved
Fine-tune SigLIP predictor (KVTP) instead of using GPT assigned scores (ablation)KVTP pruning rates · VideoMMEaccuracy: 62.83 -> 63.29Improved
Fine-tune SigLIP predictor instead of fine-tuning backbone and projector (KeyVideoLLM-style) (ablation)KVTP pruning rates · VideoMMEaccuracy: 61.23 -> 63.29Improved

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