It is very difficult to detect multi-scale synthetic aperture radar (SAR) ships, especially under complex backgrounds. Traditional constant false alarm rate methods are cumbersome in manual design and weak in migration capabilities. Based on deep learning, researchers have introduced methods that have shown good performance in order to get better detection results. However, the majority of these methods have a huge network structure and many parameters which greatly restrict the application and promotion. In this paper, a fast and lightweight detection network, namely FASC-Net, is proposed for multi-scale SAR ship detection under complex backgrounds. The proposed FASC-Net is mainly composed of ASIR-Block, Focus-Block, SPP-Block, and CAPE-Block. Specifically, without losing information, Focus-Block is placed at the forefront of FASC-Net for the first down-sampling of input SAR images at first. Then, ASIR-Block continues to down-sample the feature maps and use a small number of parameters for feature extraction. After that, the receptive field of the feature maps is increased by SPP-Block, and then CAPE-Block is used to perform feature fusion and predict targets of different scales on different feature maps. Based on this, a novel loss function is designed in the present paper in order to train the FASC-Net. The detection performance and generalization ability of FASC-Net have been demonstrated by a series of comparative experiments on the SSDD dataset, SAR-Ship-Dataset, and HRSID dataset, from which it is obvious that FASC-Net has outstanding detection performance on the three datasets and is superior to the existing excellent ship detection methods.

Automatic target recognition (ATR) in synthetic aperture radar (SAR) images has been widely used in civilian and military fields. Traditional model-based methods and template matching methods do not work well under extended operating conditions (EOCs), such as depression angle variant, configuration variant, and noise corruption. To improve the recognition performance, methods based on convolutional neural networks (CNN) have been introduced to solve such problems and have shown outstanding performance. However, most of these methods rely on continuously increasing the width and depth of networks. This adds a large number of parameters and computational overhead, which is not conducive to deployment on edge devices. To solve these problems, a novel lightweight fully convolutional neural network based on Channel-Attention mechanism, Channel-Shuffle mechanism, and Inverted-Residual block, namely the ASIR-Net, is proposed in this paper. Specifically, we deploy Inverted-Residual blocks to extract features in high-dimensional space with fewer parameters and design a Channel-Attention mechanism to distribute different weights to different channels. Then, in order to increase the exchange of information between channels, we introduce the Channel-Shuffle mechanism into the Inverted-Residual block. Finally, to alleviate the matter of the scarcity of SAR images and strengthen the generalization performance of the network, four approaches of data augmentation are proposed. The effect and generalization performance of the proposed ASIR-Net have been proved by a lot of experiments under both SOC and EOCs on the MSTAR dataset. The experimental results indicate that ASIR-Net achieves higher recognition accuracy rates under both SOC and EOCs, which is better than the existing excellent ATR methods.