UniEvo-RS: Omni-Prompt Unified Remote Sensing Segmentation with Representative Exemplar-Driven Prototype Evolution

2026-08-04

AI Summary

This work addresses the limited generalization of existing static vision-language models to novel scenes and unseen categories in remote sensing image segmentation, as well as their lack of flexible task-routing mechanisms for multi-intent annotation workflows. To overcome these challenges, the authors construct a multi-instruction prompting dataset that unifies textual and visual prompts within a single architecture and introduce a training-free prototype evolution mechanism. This mechanism generates positive and negative prototypes from representative sample feedback, leveraging contrastive learning to enhance query recall and suppress background noise. Additionally, a dynamic task routing strategy is proposed to continuously refine segmentation performance on unseen categories through minimal-sample interactions under a fixed annotation budget. Experiments demonstrate that the method significantly outperforms existing approaches across diverse remote sensing segmentation tasks, enabling efficient and accurate zero-shot incremental annotation.

Abstract

Prompt-driven vision-language models (VLMs) hold immense promise for accelerating dense remote sensing (RS) annotation, but static models suffer from severe performance degradation when deployed on novel scenes, unseen categories, or visually confusing backgrounds. Moreover, existing unified paradigms primarily rely on intra-image specific prompts, lacking flexible task routing to adapt to multi-intent operational workflows. In practical batch mapping, annotators typically refine a small set of representative samples before processing large datasets. Motivated by this practice, we propose UniEvo-RS, an omni-prompt unified RS segmentation framework equipped with representative exemplar-driven prototype evolution. First, we construct a multi-instruction prompt dataset that unifies text-driven and visual-driven prompts within a single architecture, establishing a dynamic task-routing mechanism for highly diverse RS annotation scenarios. Second, we introduce a representative feedback-driven, training-free prototype evolution mechanism. By contrasting manual annotations with initial predictions on exemplars, UniEvo-RS distills prediction errors into positive and negative prototypes. These prototypes enhance LLM query recall and suppress spatial background noise under a fixed-budget clustering memory. Extensive experiments show that UniEvo-RS unifies diverse prompting tasks, achieving state-of-the-art performance across most settings. Crucially, with minimal interaction on a few exemplars, it enables training-free, progressive accuracy enhancement on unseen categories during batch annotation.

Problem

• remote sensing segmentation
• prompt-driven models
• unseen categories
• task routing
• visual confusion

Innovation

• prototype evolution
• omni-prompt
• representative exemplar
• unified segmentation
• training-free adaptation