Abstract
In high-resolution Synthetic Aperture Radar (SAR) ship detection, the number of SAR samples seriously affects the performance of the algorithms based on deep learning. In this paper, aiming at the application requirements of high-resolution ship detection in small samples, a high-resolution SAR ship detection method combining an improved sample generation network, Multiscale Wasserstein Auxiliary Classifier Generative Adversarial Networks (MW-ACGAN) and the Yolo v3 network is proposed. Firstly, the multi-scale Wasserstein distance and gradient penalty loss are used to improve the original Auxiliary Classifier Generative Adversarial Networks (ACGAN), so that the improved network can stably generate high-resolution SAR ship images. Secondly, the multi-scale loss term is added to the network, so the multi-scale image output layers are added, and multi-scale SAR ship images can be generated. Then, the original ship data set and the generated data are combined into a composite data set to train the Yolo v3 target detection network, so as to solve the problem of low detection accuracy under small sample data set. The experimental results of Gaofen-3 (GF-3) 3 m SAR data show that the MW-ACGAN network can generate multi-scale and multi-class ship slices, and the confidence level of ResNet18 is higher than that of ACGAN network, with an average score of 0.91. The detection results of Yolo v3 network model show that the detection accuracy trained by the composite data set is as high as 94%, which is far better than that trained only by the original SAR data set. These results show that our method can make the best use of the original data set, improve the accuracy of ship detection.
Highlights
The Yolo v3 [48,49,50,51,52,53] network was selected to test whether the Synthetic aperture radar (SAR) samples generated by our
In order to meet the application requirements of ship detection in high-resolution SAR images with small samples, we propose a ship detection method based on the SAR image generation network (MW-Auxiliary Classifier Generative Adversarial Networks (ACGAN)) and the detection network Yolo v3 under small samples
The experimental results of GF-3 SAR data show that the MW-ACGAN network can effectively generate realistic multiclass ship images by using the Wasserstein distance and gradient penalty
Summary
Ship detection is an important maritime management technology, including the investigation of illegal fishing areas, oil spill detection, maritime traffic management, and national defense [1,2,3,4,5,6,7,8,9]. The calculation processes for these methods are complex, consume much computational memory, and cost a large amount of resources [35] These methods lack real data information in the simulation process, resulting in generated images with insufficient realism. The first issue is how to use GANs to generate high-resolution SAR images of multiple scales stably to solve the bottleneck problem of insufficient SAR data. This paper proposes an integrated framework of sample generation and detection in the case of small samples of high-resolution SAR ships. ACGAN method called Multiscale Wasserstein Auxiliary Classifier Generative Adversarial Networks (MW-ACGAN) is proposed, which uses the multi-scale Wasserstein distance and gradient penalty to make the network more stable, generating multiscale high-resolution SAR ship images.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
