An Enhanced RT-DETR with Dual Convolutional Kernels for SAR Ship Detection

In recent years, ship detection based on remote sensing images has emerged as a crucial task for coastal nations due to the advancement of remote sensing technology. Among active imaging sensors in remote sensing, synthetic aperture radar (SAR) stands out as one of the most significant ones because of its immunity to cloud cover and ability to operate day and night. However, ship targets in SAR images pose challenges such as indistinct contour information, intricate backgrounds, and intense scattering. Despite commendable results achieved by ship detection algorithms based on deep learning neural networks, they still suffer from numerous missed detections and false alarms. In this study, we propose an enhanced real-time detection transformer (RT -DETR) with dual convolutional kernels (DualConv) for accurate ship detection in SAR images. Numerical experiments conducted on the high-resolution SAR image dataset (HRSID) demonstrate the effectiveness of the proposed method, improving detection accuracy and model's robustness and capability in complex marine environments.