UrbanCross: Enhancing Satellite Image-Text Retrieval with Cross-Domain Adaptation

TL;DR

UrbanCross introduces a cross-domain satellite image-text retrieval framework that leverages multimodal models and adaptive training to improve retrieval accuracy across diverse urban landscapes, addressing domain gap challenges.

Contribution

The paper presents UrbanCross, a novel framework combining multimodal models and adaptive training for effective cross-domain satellite image-text retrieval in urban environments.

Findings

Achieves 10% improvement in retrieval performance.

Demonstrates 15% average performance increase with domain adaptation.

Effectively bridges domain gaps across diverse urban landscapes.

Abstract

Urbanization challenges underscore the necessity for effective satellite image-text retrieval methods to swiftly access specific information enriched with geographic semantics for urban applications. However, existing methods often overlook significant domain gaps across diverse urban landscapes, primarily focusing on enhancing retrieval performance within single domains. To tackle this issue, we present UrbanCross, a new framework for cross-domain satellite image-text retrieval. UrbanCross leverages a high-quality, cross-domain dataset enriched with extensive geo-tags from three countries to highlight domain diversity. It employs the Large Multimodal Model (LMM) for textual refinement and the Segment Anything Model (SAM) for visual augmentation, achieving a fine-grained alignment of images, segments and texts, yielding a 10% improvement in retrieval performance. Additionally, UrbanCross incorporates an adaptive curriculum-based source sampler and a weighted adversarial cross-domain fine-tuning module, progressively enhancing adaptability across various domains. Extensive experiments confirm UrbanCross's superior efficiency in retrieval and adaptation to new urban environments, demonstrating an average performance increase of 15% over its version without domain adaptation mechanisms, effectively bridging the domain gap.