Multitask Learning for Ship Detection From Synthetic Aperture Radar Images

Abstract

Ship detection from SAR images is inherently subject to the special imaging mechanism of SAR. In recent years, deeplearning-based techniques for detecting objects from optical images have rapidly advanced and promoted the development of SAR image detection technology. However, the strong speckle noise in SAR images degrades low-level feature learning in shallow layers, hindering the higher-level learning of semantic features for object detection. In view of the problems encountered in direct end-to-end feature learning for object detection and the close relationship between objects and auxiliary cues, a multitask learning-based object detector (MTL-Det) is proposed in this paper to distinguish ships in SAR images. The proposed approach models the ship detection problem, not as a single object detection task, but as three cooperative tasks. The model involves two auxiliary sub-tasks that are focused on learning object-specific cues (e.g. texture and shape) for the ship detection task, which is constrained by the pseudo-ground truth generated by the main task. Assisted by auxiliary sub-tasks, the lowlevel features are robust to speckle noise and reliably support high-level feature learning. Compared with traditional singletask-based object detectors, more discriminative object-specific features are learned by multi-task learning without the extra cost of manual labeling. The experiments conducted in this study help demonstrate the advantages of MTL-Det in improving ship detection performance on two SAR datasets: high-resolution SAR images dataset and large-scale SAR ship detection dataset-v1.0.

BiFA-YOLO: A Novel YOLO-Based Method for Arbitrary-Oriented Ship Detection in High-Resolution SAR Images

“…Hong et al [40] proposed a "you only look once" version 3 (YOLOv3) framework to detect multiscale ships from SAR and optical imagery. Zhang et al [41] proposed a multitask learning-based object detector (MTL-Det) to distinguish ships in SAR images. Li et al [42] designed a novel multidimensional domain deep learning network and exploited the spatial and frequency-domain complementary features to SAR ship detection.…”

Lite-YOLOv5: A Lightweight Deep Learning Detector for On-Board Ship Detection in Large-Scene Sentinel-1 SAR Images

“…The LS-SSDD-v1.0 dataset is widely used for SAR image intelligent interpretation [51][52][53][54]. The characteristic of small ships with large-scale backgrounds in LS-SSDD is close to actual satellite images; thus, we adopted the LS-SSDD-v1.0 dataset to verify the effectiveness of Lite-YOLOv5.…”

A Ship Detection and Imagery Scheme for Airborne Single-Channel SAR in Coastal Regions

“…Although multi-channels SAR systems improve the detectability of ship targets, singlechannel SAR systems are often equipped with the consideration of cost saving, weight constraint, and more flexible observation [12][13][14]. In recent years, some ship detection algorithms from SAR imagery based on deep learning have been widely applied [15][16][17][18][19][20][21][22]. Deep learning is a trend of ship detection from SAR images in the future, and has made significant progress, but it usually conforms to the optical ship detector (OSD) and the postprocessing of spaceborne SAR images, and depends on a high resolution or on complex computation, which makes it not suitable for a real-time SAR system such as airborne SAR [23][24][25].…”