GeoMamba: A Geometry-driven MambaVision Framework and Dataset for Fine-grained Optical-SAR Object Retrieval
Researchers have introduced GeoMamba, a novel framework designed to improve the retrieval of objects across optical and Synthetic Aperture Radar (SAR) satellite imagery, even when the images are not aligned. The framework utilizes a Geometric Feature Injection module to enhance cross-modal feature interaction and a Geometric Consistency Constraint module to preserve object structures. A new dataset, FGOS-as, was also created to evaluate this approach, with GeoMamba achieving a 63.3% mAP and 77.0% Rank-1 accuracy on fine-grained retrieval tasks. AI
IMPACT Introduces a new method for improving cross-modal satellite image analysis, potentially aiding in more robust object detection and retrieval tasks.