Abstract:Ship detection using synthetic aperture radar (SAR) images is widely applied to marine monitoring, ship identification, and other intelligent maritime applications. It also improves shipping efficiency, reduces marine traffic accidents, and promotes marine resource development. Land reflection and sea clutter introduce noise into SAR imaging, making the ship features in the image less prominent, which makes the detection of multi-scale ship targets more difficult. Therefore, a cross-scale ship detection network for SAR images based on efficient receptive field and enhanced hierarchical fusion is proposed. In order to retain more information and lighten the weight of the network, an efficient receptive field feature extraction backbone network (ERFBNet) is designed, and the multi-channel coordinate attention mechanism (MCCA) is embedded to highlight the ship features. Then, an enhanced hierarchical feature fusion network (EHFNet) is proposed to better characterize the features by fusing information from lower and higher layers. Finally, the feature map is input into the detection head with improved bounding box loss function. Using SSDD and HRSID as experimental datasets, average accuracies of 97.3% and 90.6% were obtained, respectively, and the network performed well in most scenarios.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenge of detecting multi - scale ship targets in Synthetic Aperture Radar (SAR) images. Specifically, the paper points out: 1. **Background noise problem**: Land reflection and sea clutter will introduce noise into SAR imaging, resulting in indistinct ship features in the image, which makes the detection of multi - scale ship targets more difficult. 2. **Multi - scale target detection problem**: Due to the different sizes of ships, it becomes very difficult to detect ships of all scales at the same time. 3. **Detection accuracy problem in complex scenarios**: In complex marine environments, such as when ships are docked at the shore or entering and leaving the port, near - shore buildings may cause interference, making detection more difficult. Especially in the case of densely arranged ships, most models cannot fully locate the ships, resulting in more missed detections. 4. **Small - target detection problem**: Since the network performs multiple down - sampling when extracting features, the target feature information occupying a smaller pixel area is lost, which seriously reduces the detection ability of small - scale ships. Especially when there are large ships, small ships are easily misidentified as noise. To solve the above problems, the paper proposes a Cross - Scale SAR Ship Detection Network (CSEF - Net) based on efficient receptive fields and enhanced hierarchical fusion. The main contributions of this network include: 1. **Improving the accuracy of ship detection at different scales**: By designing an efficient Receptive Field Feature Extraction Backbone Network (ERFBNet), the receptive field is expanded and more effective information is retained. At the same time, an effective attention mechanism and a lightweight convolutional aggregation module are introduced. 2. **Enhancing the multi - level feature fusion ability**: An Enhanced Hierarchical Feature Fusion Network (EHFNet) is designed. Through weighted fusion and skip - layer connections, the flow of features at different scales is promoted, context information is merged, and more accurate location and semantic information are provided. 3. **Optimizing the bounding box regression loss function**: A more effective bounding box regression loss function is designed, which combines Wise - IOU and NWD metrics and helps to improve the detection accuracy of target locations and the overall detection performance of the network. Through these improvements, CSEF - Net is optimized in network structure and algorithm, aiming to improve the accuracy and robustness of ship detection in SAR images, especially in complex scenarios and multi - scale target detection.

CSEF-Net: Cross-Scale SAR Ship Detection Network Based on Efficient Receptive Field and Enhanced Hierarchical Fusion

Enhanced Cascade R-CNN for Multi-scale Object Detection in Dense Scenes from SAR Images

SEFRNet: An SAR Ship Target Detection Network With Effective Feature Representation

Ship Detection in SAR Images Based on Multi-Scale Feature Extraction and Adaptive Feature Fusion

Stepwise Attention-Guided Multiscale Fusion Network for Lightweight and High-Accurate SAR Ship Detection

BFEA: A SAR Ship Detection Model Based on Attention Mechanism and Multiscale Feature Fusion

SFFNet:A Ship Detection Method Using Scattering Feature Fusion for Sea Surface SAR Images

Sar-Shipnet: Sar-Ship Detection Neural Network via Bidirectional Coordinate Attention and Multi-Resolution Feature Fusion

SAR ship detection network based on global context and multi-scale feature enhancement

Automatic SAR Ship Detection Based on Multi-Feature Fusion Network in Spatial and Frequency Domain

HA-Net: a SAR image ship detector based on hybrid attention

An SAR Ship Object Detection Algorithm Based on Feature Information Efficient Representation Network

FESAR: SAR ship detection model based on local spatial relationship capture and fused convolutional enhancement

Sub-pixel Feature Pyramid Network for Multi-Scale Ship Detection in Synthetic Aperture Radar Remote Sensing Images

Precise and Robust Ship Detection for High-Resolution SAR Imagery Based on HR-SDNet

MAEE-Net: SAR Ship Target Detection Network Based on Multi-Input Attention and Edge Feature Enhancement

A-BFPN: An Attention-Guided Balanced Feature Pyramid Network for SAR Ship Detection

Multi Scale Ship Detection Based on Attention and Weighted Fusion Model for High Resolution SAR Images.

Attention Receptive Pyramid Network for Ship Detection in SAR Images

Contextual Region-Based Convolutional Neural Network with Multilayer Fusion for SAR Ship Detection

Ship Detection in SAR Images via Cross-Attention Mechanism