DiCoR: Decoupled Referent Disambiguation and Contour Recalibration for Efficient Referring Remote Sensing Image Segmentation

2026-08-13

AI Summary

This work addresses the limitations of existing approaches in referring expression segmentation for remote sensing images—specifically, the insufficient accuracy of joint optimization methods and the low efficiency of decoupled strategies—by proposing DiCoR, a framework that synergistically integrates the strengths of both paradigms. DiCoR first leverages language-adaptive cues to localize candidate regions and then refines the initial coarse mask through a lightweight residual module under local contour supervision. The method further incorporates a referring disambiguation guidance strategy and an efficient feature fusion mechanism. Extensive experiments demonstrate that DiCoR achieves state-of-the-art performance across RefSegRS, RRSIS-D, and RISBench benchmarks, improving mIoU and gIoU by 5.28% and 2.87%, respectively, on RefSegRS, while operating 4.7% faster than representative decoupled methods during inference.

Abstract

Referring remote sensing image segmentation (RRSIS) aims to delineate targets specified by natural language expressions in remote sensing imagery. Existing methods mainly follow joint fusion segmentation (JFS) or decoupled prompt segmentation (DPS). JFS is efficient but often suffers from limited accuracy because referent localization and mask delineation are optimized under a unified objective, whereas DPS separates localization from mask generation using spatial prompts and foundation segmenters at the cost of higher memory consumption and inference latency. To bridge this gap, we propose DiCoR, a decoupled referent disambiguation and contour recalibration framework built on an efficient JFS pipeline. DiCoR addresses two key challenges: distinguishing the correct referent from ambiguous candidates and refining coarse masks after localization. A disambiguation-aware localization guidance strategy ranks salient candidate regions with adaptive linguistic cues and injects the resulting localization prior into fused features. A lightweight contour recalibration module further predicts residual corrections to coarse logits under localized contour supervision, improving mask quality with limited computational overhead. Experiments on RefSegRS, RRSIS-D, and RISBench show that DiCoR achieves the best segmentation accuracy across all three benchmarks. On RefSegRS, it improves mIoU and gIoU by 5.28% and 2.87% over a competitive JFS method while running 4.7% faster than a representative DPS method, demonstrating a favorable accuracy-efficiency trade-off. Code is available at https://github.com/zyGao1126/DiCoR.

Problem

• referring remote sensing image segmentation
• referent disambiguation
• contour recalibration
• accuracy-efficiency trade-off
• mask refinement

Innovation

• decoupled segmentation
• referent disambiguation
• contour recalibration
• referring remote sensing image segmentation
• efficient inference