AI Paper Atlas

Research map / Video / Video-language understanding

S3-Prune: Stability-Aware Token Budgeting for Long-Form Video-Language Models

gilha lee, Seungil Lee, Hyun Kim

ECCV 2026 · Video › Video-language understanding · approach: Long video understanding

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 Kalman filter stability accumulation (ablation)S3-prune on LLaVA-OV · 4 benchmarks (10% tokens)avg score: 53.6 -> 54.4Improved
add Kalman filter stability accumulation (ablation)S3-prune on LLaVA-Video · 4 benchmarks (10% tokens)avg score: 57.3 -> 58.1Improved
add Kalman filter stability accumulation (ablation)S3-prune on Qwen2.5-VL-7B · 4 benchmarks (10% tokens)avg score: 59.6 -> 60.2Improved
add spatial uncertainty signal to segment transition signal (ablation)S3-prune Qwen3-VL-8B · MVBench+VideoMMEavg score: 59.7 -> 60.2Improved
add segment transition signal to spatial uncertainty signal (ablation)S3-prune Qwen3-VL-8B · MVBench+VideoMMEavg score: 59.9 -> 60.2Improved
add S3-prune budgeting to FastVID segmentationFastVID · 4 benchmarks (10% tokens)avg score: 59.3 -> 59.7Improved
add S3-prune budgeting to HoliTom segmentationHoliTom · 4 benchmarks (10% tokens)avg score: 57.2 -> 57.7Improved
use S3-prune pre-LLM stage instead of FastVidFastVid · MVBench+VideoMME avg (15% tokens)avg score: 61.4 -> 61.9Improved
use S3-prune instead of HoliTom at 10% retentionHoliTom · LLaVA-OV-7B 7 benchmarksavg score: 59.6 -> 59.7Improved
use S3-prune at 25% token retentionLLaVA-OV-7B · 7 benchmarksavg score: 60.7 -> 61.1Improved
use S3-prune at 25% token retentionLLaVA-Video-7B · 7 benchmarksavg score: 65.9 -> 65.8No clear effect
use S3-prune at 10% retention instead of HoliTomHoliTom · Qwen3-VL-8B 4 benchmarksavg performance retained (%): 88.1 -> 93.3Improved

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