Ship Detection in Synthetic Aperture Radar Images with Improved YOLOv8

In recent years, ship detection using remote sensing images has emerged as a crucial task for coastal countries due to advancements in remote sensing technology. Synthetic aperture radar (SAR) is a prominent active imaging sensor, unaffected by clouds and capable of day-night operation. However, SAR images pose challenges such as unclear contours, complex backgrounds, and strong scattering, leading to the misdetection and missed detection of small ship targets. To address these issues, this paper proposes an improved ship detection model for SAR images based on the YOLOv8 framework. Our approach introduces a small target detection layer to the original YOLOv8 architecture and adapt the loss function using the wise-intersection over union. Experimental evaluations conducted on the HRSID and the SSDD datasets demonstrate the effectiveness of our method, improving detection accuracy, recall, and robustness in complex marine environments.