SOPSeg: Prompt-based Small Object Instance Segmentation in Remote Sensing Imagery

2025-09-03

AI Summary

To address the lack of dedicated datasets and efficient methods for small-object instance segmentation in remote sensing imagery, this paper introduces the first prompt-based segmentation framework tailored for micro-objects. To overcome SAM’s boundary ambiguity on small objects—caused by its 1/16-resolution feature map—we propose: (i) a region-adaptive upsampling strategy; (ii) a rotated bounding-box-guided prompting mechanism; (iii) an edge-aware segmentation head; and (iv) a progressive prompt optimization decoder, collectively enhancing fine-grained boundary accuracy. We further construct SODA-S, the first benchmark dataset for remote sensing small-object instance segmentation, containing over 12K high-precision pixel-level annotations. Both the dataset and model are publicly released. Extensive experiments demonstrate substantial improvements over state-of-the-art methods across multiple remote sensing benchmarks, establishing a new paradigm and foundational resource for small-object segmentation.

Technology Category

• Computer Vision: Segmentation
• Natural Language Processing: Prompt Engineering / Prompting
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

Extracting small objects from remote sensing imagery plays a vital role in various applications, including urban planning, environmental monitoring, and disaster management. While current research primarily focuses on small object detection, instance segmentation for small objects remains underexplored, with no dedicated datasets available. This gap stems from the technical challenges and high costs of pixel-level annotation for small objects. While the Segment Anything Model (SAM) demonstrates impressive zero-shot generalization, its performance on small-object segmentation deteriorates significantly, largely due to the coarse 1/16 feature resolution that causes severe loss of fine spatial details. To this end, we propose SOPSeg, a prompt-based framework specifically designed for small object segmentation in remote sensing imagery. It incorporates a region-adaptive magnification strategy to preserve fine-grained details, and employs a customized decoder that integrates edge prediction and progressive refinement for accurate boundary delineation. Moreover, we introduce a novel prompting mechanism tailored to the oriented bounding boxes widely adopted in remote sensing applications. SOPSeg outperforms existing methods in small object segmentation and facilitates efficient dataset construction for remote sensing tasks. We further construct a comprehensive small object instance segmentation dataset based on SODA-A, and will release both the model and dataset to support future research.

Problem

• Addresses small object instance segmentation in remote sensing imagery
• Overcomes SAM's feature resolution limitations for fine spatial details
• Solves lack of dedicated datasets and high annotation costs

Innovation

• Region-adaptive magnification strategy preserves fine details
• Customized decoder integrates edge prediction and refinement
• Novel prompting mechanism for oriented bounding boxes