Abstract

Recent years have witnessed great progress in synthetic aperture radar (SAR) target detection methods based on deep learning. However, these methods generally assume the training data and test data obey the same distribution, which does not always hold when the radar parameters, imaging algorithm, viewpoints, scenes, etc., change in practice. When such a distribution mismatch occurs, it will cause a significant performance drop. Domain adaptation methods provide an effective way to address this problem by transferring knowledge from the source domain (training data) to the target domain (test data). In this article, we proposed an unsupervised faster R-CNN SAR target detection framework based on domain adaptation, which can improve SAR target detection performance in the unlabeled target domain by borrowing the knowledge of the labeled source domain. Our approach is composed of the following three stages: pixel-domain adaptation (PDA), multilevel feature domain adaptation (MFDA), and iterative pseudolabeling (IPL). By generating transition domain using generative adversarial networks, the PDA stage can reduce the appearance differences of SAR images. At the MFDA stage, the detector can not only learn the domain-invariant global features and instance-level regional features via multilevel adversarial learning in the common feature space but also reweight the low-level global features according to their relative importance to the target domain. At the IPL stage, we design an iterative pseudo labeling strategy that can select pseudo-labels on instance level and image level to encourage the detector to learn more discriminative features of the target domain directly. We evaluate our method using miniSAR and FARADSAR datasets. The experimental results demonstrate the effectiveness of the proposed unsupervised domain adaptation target detection approach.

Highlights

  • W ITH the advantage of providing remote sensing images under all-day and all-weather conditions, synthetic aperture radar (SAR) is widely used in the military and civilian fields

  • We focus on unsupervised domain adaptation for Synthetic Aperture Radar (SAR) target detection, where the source domain data is fully labeled while the target domain data is unlabeled

  • 9 SAR images in the miniSAR dataset are selected for the experiment, 7 of which are used for training and 2 for testing, and 106 SAR images in the FARADSAR dataset are selected for the experiment, 78 of which are used for training and 28 for testing

Read more

Summary

Introduction

W ITH the advantage of providing remote sensing images under all-day and all-weather conditions, synthetic aperture radar (SAR) is widely used in the military and civilian fields. As a fundamental and challenging task in SAR image interpretation, SAR target detection is obtaining wide attention [2]-[8]. With the recent development of the deep learning, a large number of methods based on convolutional neural networks (CNNs) are proposed. Due to numerous training data that can be learned by the network, these methods have achieved significant progress in target detection, among which the Faster R-CNN [1] is one of the most widely used methods and can be extended to other tasks flexibly. Inspired by great success in target detection of optical images, the use of CNNs in SAR target detection [2]-[8] is obtained wide attention. Li et al [4] proposed an improved Faster R-CNN by fusing features of the last three convolution layers. Jiao et al [5] connected multi-scale feature map densely based on Faster R-CNN to solve multi-scale and multi-scene problems

Objectives
Results
Conclusion
Full Text
Published version (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