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

2026-04-13

AI Summary

This work addresses the high computational cost incurred by the massive visual tokens generated by multimodal large language models when processing ultra-high-resolution remote sensing imagery, a challenge exacerbated by existing compression methods that rely on static strategies and struggle to simultaneously preserve semantic meaning and geometric structure. To overcome this limitation, the authors propose DualComp, a novel framework featuring the first semantic-geometry dual adaptive compression mechanism. Leveraging a lightweight pretrained router for dynamic guidance, DualComp decouples feature processing into two parallel streams: a Spatially-Contiguous Semantic Aggregator (SCSA) performs size-adaptive clustering to compress background regions, while an Instruction-Guided Structure Recoverer (IGSR) reconstructs spatial skeletons via greedy path tracing. Requiring no additional training, DualComp adaptively selects compression strategies per task, achieving significant gains in both inference efficiency and accuracy on XLRS-Bench while enabling high-fidelity, low-overhead remote sensing image understanding.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Multimodal Large Language Models (MLLMs) have demonstrated immense potential in Earth observation. However, the massive visual tokens generated when processing Ultra-High-Resolution (UHR) imagery introduce prohibitive computational overhead, severely bottlenecking their inference efficiency. Existing visual token compression methods predominantly adopt static and uniform compression strategies, neglecting the inherent "Semantic-Geometric Duality" in remote sensing interpretation tasks. Specifically, object semantic tasks focus on the abstract semantics of objects and benefit from aggressive background pruning, whereas scene geometric tasks critically rely on the integrity of spatial topology. To address this challenge, we propose DualComp, a task-adaptive dual-stream token compression framework. Dynamically guided by a lightweight pre-trained router, DualComp decouples feature processing into two dedicated pathways. In the object semantic stream, the Spatially-Contiguous Semantic Aggregator (SCSA) utilizes size-adaptive clustering to aggregates redundant background while protecting small object. In the scene geometric stream, the Instruction-Guided Structure Recoverer (IGSR) introduces a greedy path-tracing topology completion mechanism to reconstruct spatial skeletons. Experiments on the UHR remote sensing benchmark XLRS-Bench demonstrate that DualComp accomplishes high-fidelity remote sensing interpretation at an exceptionally low computational cost, achieving simultaneous improvements in both efficiency and accuracy.

Problem

• Ultra-High-Resolution
• Visual Token Compression
• Semantic-Geometric Duality
• Remote Sensing Understanding
• Computational Overhead

Innovation

• Semantic-Geometric Duality
• Visual Token Compression
• Task-Adaptive Framework
• Ultra-High-Resolution Remote Sensing
• Training-Free Compression