AI Paper Atlas

Research map / Video / Video-language understanding

ST-SimDiff: Balancing Spatiotemporal Similarity and Difference for Efficient Video Understanding with MLLMs

Bingjun Luo, Tony Wang, Chaoqi Chen, Xinpeng Ding

ICLR 2026 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding · 1 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 spatial similarity selection to baseline (ablation)ST-SimDiff (LLaVA-Video-7B, retain 30%) · VideoMME overallaccuracy (%): 60.3 -> 61.5Improved
add temporal similarity selection to baseline (ablation)ST-SimDiff (LLaVA-Video-7B, retain 30%) · VideoMME overallaccuracy (%): 60.3 -> 61.7Improved
use joint spatio-temporal similarity instead of either alone (ablation)ST-SimDiff (LLaVA-Video-7B, retain 30%) · VideoMME overallaccuracy (%): 61.7 -> 62.6Improved
add difference-based event token selection (Diff) (ablation)ST-SimDiff (LLaVA-Video-7B, retain 30%) · VideoMME overallaccuracy (%): 62.6 -> 63.2Improved
add difference-based event token selection (Diff) (ablation)ST-SimDiff (LLaVA-Video-7B, retain 30%) · EgoSchemaaccuracy (%): 55.3 -> 56.0Improved
add difference-based event token selection (Diff) (ablation)ST-SimDiff (LLaVA-Video-7B, retain 50%) · VideoMME overallaccuracy (%): 63.7 -> 63.8Improved
add difference-based event token selection (Diff) (ablation)ST-SimDiff (NVILA-Video-8B, retain 30%) · VideoMME overallaccuracy (%): 59.9 -> 60.2Improved
add difference-based event token selection (Diff) (ablation)ST-SimDiff (NVILA-Video-8B, retain 50%) · LongVideoBench overallaccuracy (%): 56.0 -> 56.5Improved
add similarity modules to baseline (ablation)ST-SimDiff (NVILA-Video-8B, retain 30%) · VideoMME overallaccuracy (%): 59.3 -> 59.9Improved
increase token retain ratio from 30% to 50%ST-SimDiff (LLaVA-Video-7B) · VideoMME overallaccuracy (%): 63.2 -> 63.8Improved
increase input frames from 64 to 128ST-SimDiff (Qwen2.5-VL-7B, retain 50%) · VideoMME overallaccuracy (%): 62.9 -> 66.3Improved
compress with ST-SimDiff at 30% retention instead of keeping all tokensLLaVA-Video-7B · VideoMME overallaccuracy (%): 63.3 -> 63.2No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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