SLIP-RS: Structured-Attribute Language-Image Pre-Training for Remote Sensing Object Detection

Chenxu Wang, Yuxuan Li, Yunheng Li, Xiang Li, Jingyuan Xia, Qibin Hou

arXiv:2605.23144 · 2026-05-25 공개 · arXiv · PDF

contrastive-learning remote-sensing conformal-prediction cross-domain-generalization language-image-pre-training rs-attribute-15m fine-grained-detection structured-attribute

Abstract

Existing language-image pre-training for remote sensing object detection is constrained by Monolithic Label Learning, which relies on exhaustively enumerating open-set categories via black-box data to acquire fine-grained representations, creating a dependency incompatible with the domain's inherent data scarcity. To transcend this bottleneck, we propose SLIP-RS, establishing a Structured-Attribute Decoupling Paradigm that maps the open-ended category space into a finite, physically meaningful attribute space, unlocking fine-grained discriminability via explicit structural logic. This paradigm is realized via two technical pillars: (1) Structured-Attribute Contrastive Learning, which enforces the learning of decoupled intrinsic visual logic via combinatorial attribute augmentation; and (2) Conformal Attribute Reliability Engine, which leverages conformal prediction theory to rigorously distill high-fidelity supervision from noisy sources, yielding RS-Attribute-15M, the largest dataset with over 15 million attribute annotations. Extensive experiments demonstrate that SLIP-RS establishes unprecedented performance in fine-grained detection and cross-domain generalization, validating structured attributes as a vital foundation for remote sensing. Code: https://github.com/facias914/SLIP-RS.

한국어 요약

📋 한 줄 요약

**[원격탐사 / Vision-Language Pretraining]** SLIP-RS가 monolithic label 의존을 깨고 structured-attribute decoupling으로 카테고리 공간을 attribute 공간에 매핑, conformal reliability engine으로 15M attribute 어노테이션 RS-Attribute-15M 구축·fine-grained·cross-domain SOTA.

🎯 핵심 기여도

💡 핵심 아이디어

원격탐사처럼 데이터가 부족한 도메인에서 open-set category를 enumerate하는 monolithic approach 대신, 카테고리를 finite·physically meaningful attribute space로 decouple하면 explicit structural logic을 통해 fine-grained 변별을 얻을 수 있으며, conformal prediction으로 noisy 어노테이션 신뢰성을 보장할 수 있다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 원격탐사 vision-language pretraining의 패러다임 전환(category→attribute), conformal prediction의 데이터 큐레이션 적용 모범, 15M annotation 공개로 커뮤니티 자산. **한계**: Attribute 정의의 도메인 전문 지식 의존, structured attribute가 일반 자연 이미지 일반화는 별도, conformal reliability engine의 hyperparameter 의존.

🚀 실용적 활용