SARES-DEIM: Sparse Mixture-of-Experts Meets DETR for Robust SAR Ship Detection

TL;DR

SARES-DEIM is a novel SAR ship detection framework that combines sparse mixture-of-experts with DETR, effectively filtering noise and preserving small target details, leading to superior detection performance.

Contribution

The paper introduces SARES-DEIM, integrating SAR-aware experts and a high-resolution feature pyramid to enhance robustness and accuracy in SAR ship detection.

Findings

Achieves 76.4% mAP50:95 on HRSID dataset.

Outperforms state-of-the-art YOLO and SAR detectors.

Effectively filters speckle noise while maintaining small target localization.

Abstract

Ship detection in Synthetic Aperture Radar (SAR) imagery is fundamentally challenged by inherent coherent speckle noise, complex coastal clutter, and the prevalence of small-scale targets. Conventional detectors, primarily designed for optical imagery, often exhibit limited robustness against SAR-specific degradation and suffer from the loss of fine-grained ship signatures during spatial downsampling. To address these limitations, we propose SARES-DEIM, a domain-aware detection framework grounded in the DEtection TRansformer (DETR) paradigm. Central to our approach is SARESMoE (SAR-aware Expert Selection Mixture-of-Experts), a module leveraging a sparse gating mechanism to selectively route features toward specialized frequency and wavelet experts. This sparsely-activated architecture effectively filters speckle noise and semantic clutter while maintaining high computational efficiency. Furthermore, we introduce the Space-to-Depth Enhancement Pyramid (SDEP) neck to preserve high-resolution spatial cues from shallow stages, significantly improving the localization of small targets. Extensive experiments on two benchmark datasets demonstrate the superiority of SARES-DEIM. Notably, on the challenging HRSID dataset, our model achieves a mAP50:95 of 76.4% and a mAP50 of 93.8%, outperforming state-of-the-art YOLO-series and specialized SAR detectors.