AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Semantic-Geometric Dual Compression: Training-Free Visual Token Reduction for Ultra-High-Resolution Remote Sensing Understanding

Yueying Li, Fengxiang Wang, Yan Li, Mingshuo Chen, Mengying Zhao, Long Lan

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Attention-based visual token pruning · 2 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 geometric stream (IGSR) to semantic stream (ablation)DualComp · XLRS-Benchavg accuracy (%): 50.3 -> 53.1Improved
add semantic stream (SCSA) to geometric stream (ablation)DualComp · XLRS-Benchavg accuracy (%): 51.1 -> 53.1Improved
add explicit topology completion (path tracing) over Top-K selection (ablation)DualComp · XLRS-Benchavg accuracy (%): 50.5 -> 53.1Improved
add topological sequence unrolling over index reordering (ablation)DualComp · XLRS-Benchavg accuracy (%): 51.5 -> 53.1Improved
add text-aware structural modulation (ablation)DualComp · XLRS-Benchavg accuracy (%): 51.2 -> 53.1Improved
add geometric stream IGSR (ablation)DualComp · XLRS-Bench Route Planningaccuracy (%): 67.0 -> 72.0Improved
add text-aware modulation (ablation)DualComp · XLRS-Bench OLUCaccuracy (%): 48.0 -> 53.0Improved
use DualComp instead of GeoLLaVA-8K static compressionGeoLLaVA-8K · XLRS-Benchavg accuracy (%): 51.5 -> 53.1Improved
use DualComp instead of VisionZip at 24xGeoLLaVA-8K · XLRS-Benchavg accuracy (%): 42.8 -> 53.1Improved
use DualComp at 42.4x instead of 24x uncompressed baseline speedGeoLLaVA-8K · Efficiencyinference speed (s/image): 8.15 -> 3.87Improved
use DualComp to cut LLM computeGeoLLaVA-8K · EfficiencyLLM TFLOPs: 198.1 -> 99.8Improved
apply DualComp to general-purpose Qwen2.5-VL-7BQwen2.5-VL-7B · XLRS-Bench (Qwen2.5-VL-7B)avg accuracy (%): 47.4 -> 47.9Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 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