Synthetic Aperture Radar Image Data Sets Research Articles

YOLOv8-BYTE: Ship tracking algorithm using short-time sequence SAR images for disaster response leveraging GeoAI

Ship tracking technology is crucial for emergency rescue in the event of a disaster. Quickly identifying the position and status of vessels is vital for rescue teams to be able to deploy efficiently in disaster areas. When responding to emergencies or natural disasters, ship tracking technology plays a critical role in supporting emergency rescue operations and resource allocation, improving the overall resilience of the maritime transportation system. However, the research on multi-object tracking (MOT) algorithms has primarily focused on optical image datasets. In contrast, image data from synthetic aperture radar (SAR) presents unique challenges, such as defocus interference, a high false alarm rate, and a lack of prior samples. To overcome these particular challenges, we propose a robust MOT algorithm developed for SAR images to achieve effective multi-vessel tracking under difficult imaging conditions. In particular, we optimize the YOLOv8 detection network by introducing a diffusion model-based training method for data augmentation. This method improves the robustness of the network to scaling, rotational and translational deformations. Moreover, an enhanced swin transformer is proposed as a feature extraction network, which strengthens the representation capability of the detection network. Furthermore, the state parameters within the KF technique are enhanced by directly capturing the details of the height and width of the tracking rectangle box. This refinement of the ByteTrack algorithm aims to achieve a more precise and accurate fit of the tracking rectangle to the ship, further improving the overall tracking performance. The experimental results from the ship detection and multiple objects tracking datasets show the impressive performance of the proposed model. With a precision of 97.60%, a recall of 96.36%, and an average precision of 96.72%, the model achieves exceptional detection accuracy with an 18% reduction in model parameters. Furthermore, significant improvements can be observed in key tracking metrics such as HOTA, MOTA and IDF1, with improvements of 4.8%, 8.5% and 6.8% respectively compared to the baseline algorithm, alongside a remarkable 37.5% reduction in IDS. It is noteworthy that the tracker works in real time, achieving an average analysis speed of 47 frames per second. The proposed MOT algorithm achieves state-of-art tracking performance on a SAR image dataset with short time sequences. Therefore, the proposed approach is a compelling solution for ship tracking in SAR imagery.

Object-Enhanced YOLO Networks for Synthetic Aperture Radar Ship Detection

Synthetic aperture radar (SAR) enables precise object localization and imaging, which has propelled the rapid development of algorithms for maritime ship identification and detection. However, most current deep learning-based algorithms tend to increase network depth to improve detection accuracy, which may result in the loss of effective features of the target. In response to this challenge, this paper innovatively proposes an object-enhanced network, OE-YOLO, designed specifically for SAR ship detection. Firstly, we input the original image into an improved CFAR detector, which enhances the network’s ability to localize and perform object extraction by providing more information through an additional channel. Additionally, the Coordinate Attention mechanism (CA) is introduced into the backbone of YOLOv7-tiny to improve the model’s ability to capture spatial and positional information in the image, thereby alleviating the problem of losing the position of small objects. Furthermore, to enhance the model’s detection capability for multi-scale objects, we optimize the neck part of the original model to integrate the Asymptotic Feature Fusion (AFF) network. Finally, the proposed network model is thoroughly tested and evaluated using publicly available SAR image datasets, including the SAR-Ship-Dataset and HRSID dataset. In comparison to the baseline method YOLOv7-tiny, OE-YOLO exhibits superior performance with a lower parameter count. When compared with other commonly used deep learning-based detection methods, OE-YOLO demonstrates optimal performance and more accurate detection results.

Multi-Level Feature-Refinement Anchor-Free Framework with Consistent Label-Assignment Mechanism for Ship Detection in SAR Imagery

Deep learning-based ship-detection methods have recently achieved impressive results in the synthetic aperture radar (SAR) community. However, numerous challenging issues affecting ship detection, such as multi-scale characteristics of the ship, clutter interference, and densely arranged ships in complex inshore, have not been well solved so far. Therefore, this article puts forward a novel SAR ship-detection method called multi-level feature-refinement anchor-free framework with a consistent label-assignment mechanism, which is capable of boosting ship-detection performance in complex scenes. First, considering that SAR ship detection is susceptible to complex background interference, we develop a stepwise feature-refinement backbone network to refine the position and contour of the ship object. Next, we devise an adjacent feature-refined pyramid network following the backbone network. The adjacent feature-refined pyramid network consists of the sub-pixel sampling-based adjacent feature-fusion sub-module and adjacent feature-localization enhancement sub-module, which can improve the detection capability of multi-scale objects by mitigating multi-scale high-level semantic loss and enhancing low-level localization features. Finally, to solve the problems of unbalanced positive and negative samples and densely arranged ship detection, we propose a consistent label-assignment mechanism based on consistent feature scale constraints to assign more appropriate and consistent labels to samples. Extensive qualitative and quantitative experiments on three public datasets, i.e., SAR Ship-Detection Dataset (SSDD), High-Resolution SAR Image Dataset (HRSID), and SAR-Ship-Dataset illustrate that the proposed method is superior to many state-of-the-art SAR ship-detection methods.

An efficient YOLO for ship detection in SAR images via channel shuffled reparameterized convolution blocks and dynamic head

Synthetic aperture radar (SAR) is a crucial active imaging technology in remote sensing, offering valuable information for applications like climate monitoring, environmental analysis, and ship surveillance. Ship detection in SAR images remains challenging due to diverse vessel types and environmental interference, especially in inshore areas, despite the proven effectiveness of deep learning-based algorithms. This paper presents an efficient deep learning method named you only look once-shuffle reparameterized blocks with dynamic head (YOLO-SRBD) based on YOLOv8. Additionally, post-processing incorporates the soft non-maximum suppression to enhance precision. Experiments conducted on SAR image datasets demonstrate that the proposed method surpasses the original YOLOv8 both qualitatively and quantitatively, highlighting its feasibility for practical applications. The detection accuracy of the proposed YOLO-SRBD in the high resolution SAR images dataset rose from 89.9% to 91.3%, and the average precision increased from 66.7% to 74.3%, showing significant performance enhancement.

Enhanced Detection Method for Small and Occluded Targets in Large-Scene Synthetic Aperture Radar Images

Ship detection in large-scene offshore synthetic aperture radar (SAR) images is crucial in civil and military fields, such as maritime management and wartime reconnaissance. However, the problems of low detection rates, high false alarm rates, and high missed detection rates of offshore ship targets in large-scene SAR images are due to the occlusion of objects or mutual occlusion among targets, especially for small ship targets. To solve this problem, this study proposes a target detection model (TAC_CSAC_Net) that incorporates a multi-attention mechanism for detecting marine vessels in large-scene SAR images. Experiments were conducted on two public datasets, the SAR-Ship-Dataset and high-resolution SAR image dataset (HRSID), with multiple scenes and multiple sizes, and the results showed that the proposed TAC_CSAC_Net model achieves good performance for both small and occluded target detection. Experiments were conducted on a real large-scene dataset, LS-SSDD, to obtain the detection results of subgraphs of the same scene. Quantitative comparisons were made with classical and recently developed deep learning models, and the experiments demonstrated that the proposed model outperformed other models for large-scene SAR image target detection.

A Fast Facet-Based SAR Imaging Model and Target Detection Based on YOLOv5 with CBAM and Another Detection Head

Synthetic Aperture Radar (SAR) image target detection is of great significance in civil surveillance and military reconnaissance. However, there are few publicly released SAR image datasets of typical non-cooperative targets. Aiming to solve this problem, a fast facet-based SAR imaging model is proposed to simulate the SAR images of non-cooperative aircraft targets under different conditions. Combining the iterative physical optics and the Kirchhoff approximation, the scattering coefficient of each facet on the target and rough surface can be obtained. Then, the radar echo signal of an aircraft target above a rough surface environment can be generated, and the SAR images can be simulated under different conditions. Finally, through the simulation experiments, a dataset of typical non-cooperative targets can be established. Combining the YOLOv5 network with the convolutional block attention module (CBAM) and another detection head, a SAR image target detection model based on the established dataset is realized. Compared with other YOLO series detectors, the simulation results show a significant improvement in precision. Moreover, the automatic target recognition system presented in this paper can provide a reference for the detection and recognition of non-cooperative aircraft targets and has great practical application in situational awareness of battlefield conditions.

A Marine Small-Targets Classification Algorithm Based on Improved Convolutional Neural Networks

Deep learning, especially convolutional neural network (CNN) techniques, has been shown to have superior performance in ship classification, as have small-target recognition studies in safety inspections of hydraulic structures such as ports and dams. High-resolution synthetic aperture radar (SAR)-based maritime ship classification plays an increasingly important role in marine surveillance, marine rescue, and maritime ship management. To improve ship classification accuracy and training efficiency, we proposed a CNN-based ship classification method. Firstly, the image characteristics of different ship structures and the materials of ship SAR images were analyzed. We then constructed a ship SAR image dataset and performed preprocessing operations such as averaging. Combined with a classic neural network structure, we created a new convolutional module, namely, the Inception-Residual Controller (IRC) module. A convolutional neural network was built based on the IRC module to extract image features and establish a ship classification model. Finally, we conducted simulation experiments for ship classification and analyzed the experimental results for comparison. The experimental results showed that the average accuracy of ship classification of the model in this paper reached 98.71%, which was approximately 3% more accurate than the traditional network model and approximately 1% more accurate compared with other recently improved models. The new module also performed well in evaluation metrics, such as the recall rate, with accurate classifications. The model could satisfactorily describe different ship types. Therefore, it could be applied to marine ship classification management with the possibility of being extended to hydraulic building target recognition tasks.

SIVED: A SAR Image Dataset for Vehicle Detection Based on Rotatable Bounding Box

The research and development of deep learning methods are heavily reliant on large datasets, and there is currently a lack of scene-rich datasets for synthetic aperture radar (SAR) image vehicle detection. To address this issue and promote the development of SAR vehicle detection algorithms, we constructed the SAR Image dataset for VEhicle Detection (SIVED) using Ka, Ku, and X bands of data. Rotatable bounding box annotations were employed to improve positioning accuracy, and an algorithm for automatic annotation was proposed to improve efficiency. The dataset exhibits three crucial properties: richness, stability, and challenge. It comprises 1044 chips and 12,013 vehicle instances, most of which are situated in complex backgrounds. To construct a baseline, eight detection algorithms are evaluated on SIVED. The experimental results show that all detectors achieved high mean average precision (mAP) on the test set, highlighting the dataset’s stability. However, there is still room for improvement in the accuracy with respect to the complexity of the background. In summary, SIVED fills the gap in SAR image vehicle detection datasets and demonstrates good adaptability for the development of deep learning algorithms.

Cross-scene target detection based on feature adaptation and uncertainty-aware pseudo-label learning for high resolution SAR images

The characteristics of synthetic aperture radar (SAR) images are easily affected by factors such as sensor parameters, imaging scenes, etc., which may lead to data distributional discrepancies between the ideal training phase and the practical application phase, resulting in generalization performance degradation of deep learning based detectors in practice. In this paper, an accurate and fast domain adaptive detection method is proposed to address the performance degradation problem of cross-scene detection without requiring additional manual annotations. The cross-scene detection performance is gradually improved by confidence-induced unsupervised feature adaptation (CUFA) and uncertainty-aware adaptive pseudo-label learning (UAPL). In CUFA, image-level and instance-level feature adaptation are progressively implemented to bridge the distribution discrepancies between the source and the target domain in the latent feature space. Then, UAPL is performed to adaptively introduce accurate and stable supervisions of target domain according to the learning status of the detector, avoiding error accumulations and further improving the cross-scene target detection performance. The proposed cross-scene detection method is validated on airborne and spaceborne SAR image datasets of different imaging scenes. The experimental results show that by combining both CUFA and UAPL, compared with the second-best comparison method, the proposed method significantly improves the cross-scene target detection performance in SAR images (at least + 6.5% AP50 and F1 in miniSAR and FARAD cross-scene tasks, and at least + 7.5% AP50 and F1 in Gaofen-3 and Sentinel-1 cross-scene tasks), which facilitates application of the deep learning based detectors in the realistic scenarios.

Two-Way Generation of High-Resolution EO and SAR Images via Dual Distortion-Adaptive GANs

Synthetic aperture radar (SAR) provides an all-weather and all-time imaging platform, which is more reliable than electro-optical (EO) remote sensing imagery under extreme weather/lighting conditions. While many large-scale EO-based remote sensing datasets have been released for computer vision tasks, there are few publicly available SAR image datasets due to the high costs associated with acquisition and labeling. Recent works have applied deep learning methods for image translation between SAR and EO. However, the effectiveness of those techniques on high-resolution images has been hindered by a common limitation. Non-linear geometric distortions, induced by different imaging principles of optical and radar sensors, have caused insufficient pixel-wise correspondence between an EO-SAR patch pair. Such a phenomenon is not prominent in low-resolution EO-SAR datasets, e.g., SEN1-2, one of the most frequently used datasets, and thus has been seldom discussed. To address this issue, a new dataset SN6-SAROPT with sub-meter resolution is introduced, and a novel image translation algorithm designed to tackle geometric distortions adaptively is proposed in this paper. Extensive experiments have been conducted to evaluate the proposed algorithm, and the results have validated its superiority over other methods for both SAR to EO (S2E) and EO to SAR (E2S) tasks, especially for urban areas in high-resolution images.

Arbitrary-Oriented Ship Detection Method Based on Long-Edge Decomposition Rotated Bounding Box Encoding in SAR Images

Due to the limitations of the horizontal bounding boxes for locating the oriented ship targets in synthetic aperture radar (SAR) images, the rotated bounding box (RBB) has received wider attention in recent years. First, the existing RBB encodings suffer from boundary discontinuity problems, which interfere with the convergence of the model, and then lead to some problems, such as the inaccurate location of the ship targets in the boundary state. Thus, from the perspective that the long-edge features of the ships are more representative of their orientation, the long-edge decomposition RBB encoding has been proposed in this paper, which can avoid the boundary discontinuity problem. Second, the problem of the positive and negative samples imbalance is serious for the SAR ship images because only a few ship targets exist in the vast background of these images. Since the ship targets of different sizes are subject to varying degrees of interference caused by this problem, a multiscale elliptical Gaussian sample balancing strategy has been proposed in this paper, which can mitigate the impact of this problem by labeling the loss weights of the negative samples within the target foreground area with multiscale elliptical Gaussian kernels. Finally, experiments based on the CenterNet model were implemented on the benchmark SAR image dataset SSDD (SAR ship detection dataset). The experimental results demonstrate that our proposed long-edge decomposition RBB encoding outperforms other conventional RBB encodings in the task of oriented ship detection in SAR images. In addition, our proposed multiscale elliptical Gaussian sample balancing strategy is effective and can improve the model performance.

Contrastive Domain Adaptation-Based Sparse SAR Target Classification under Few-Shot Cases

Due to the imaging mechanism of synthetic aperture radar (SAR), it is difficult and costly to acquire abundant labeled SAR images. Moreover, a typical matched filtering (MF) based image faces the problems of serious noise, sidelobes, and clutters, which will bring down the accuracy of SAR target classification. Different from the MF-based result, a sparse image shows better quality with less noise and higher image signal-to-noise ratio (SNR). Therefore, theoretically using it for target classification will achieve better performance. In this paper, a novel contrastive domain adaptation (CDA) based sparse SAR target classification method is proposed to solve the problem of insufficient samples. In the proposed method, we firstly construct a sparse SAR image dataset by using the complex image based iterative soft thresholding (BiIST) algorithm. Then, the simulated and real SAR datasets are simultaneously sent into an unsupervised domain adaptation framework to reduce the distribution difference and obtain the reconstructed simulated SAR images for subsequent target classification. Finally, the reconstructed simulated images are manually labeled and fed into a shallow convolutional neural network (CNN) for target classification along with a small number of real sparse SAR images. Since the current definition of the number of small samples is still vague and inconsistent, this paper defines few-shot as less than 20 per class. Experimental results based on MSTAR under standard operating conditions (SOC) and extended operating conditions (EOC) show that the reconstructed simulated SAR dataset makes up for the insufficient information from limited real data. Compared with other typical deep learning methods based on limited samples, our method is able to achieve higher accuracy especially under the conditions of few shots.

Robust Multi-Ship Tracker in SAR Imagery by Fusing Feature Matching and Modified KCF

In previous research, most Multi-object tracking (MOT) algorithms focus on the optical image dataset, while the Synthetic Aperture Radar (SAR) image dataset faces the characteristics of few prior samples, high false alarm rate, and various defocusing interference. On the SAR image dataset, a robust MOT algorithm is proposed to fulfill multi-ship tracking in complex imaging conditions. First, the kernelized correlation filters (KCF) algorithm, a single-object tracking algorithm, is modified and applied to reduce the impact of false alarms on tracking performance. After that, different matching strategies are adaptively adapted to associate the targets based on the three intersection patterns between the predictions and the detections, which can reduce the impact of the deviated detections. Finally, the tracker’s time limit with Gaussian distribution is proposed to improve the re-association ability after the tracking interruption caused by the defocusing. The experiment results demonstrate the robust tracking ability of the proposed MOT algorithm.

SAR Image Compression Based on Multi-Resblock and Global Context

The synthetic aperture radar(SAR) image is widely used in many remote sensing applications. In order to store and transmit the increasing SAR image data, more efficient compression algorithms are needed. The purpose of this paper is to introduce a new framework for compressing SAR images. Firstly, we propose a novel analysis and synthesis transform based on multi-Resblocks for transforming the original SAR image into a compact latent representation. Then, a Gaussian mixture model is used to estimate the latent representation’s distribution. In order to explore the redundancy within the latent representation, the entropy model parameter is estimated by combining the local context, global context, and hyperprior information. In order to evaluate the performance of the proposed algorithm, we conduct experiments on a dataset of SAR images. The results show that the proposed algorithm outperforms JPEG2000 and some state-of-the-art learned image compression schemes in terms of compression performance.

SSN: Stockwell Scattering Network for SAR Image Change Detection

Recently, synthetic aperture radar (SAR) image change detection has become an interesting yet challenging direction due to the presence of speckle noise. Although both traditional and modern learning-driven methods attempted to overcome this challenge, deep convolutional neural networks (DCNNs)-based methods are still hindered by the lack of interpretability and the requirement of large computation power. To overcome this drawback, wavelet scattering network (WSN) and Fourier scattering network (FSN) are proposed. Combining respective merits of WSN and FSN, we propose Stockwell scattering network (SSN) based on Stockwell transform (ST), which is widely applied against noisy signals and shows advantageous characteristics in speckle reduction. The proposed SSN provides noise-resilient feature representation and obtains state-of-the-art performance in SAR image change detection as well as high computational efficiency. Experimental results on three real SAR image datasets demonstrate the effectiveness of the proposed method.

HRLE-SARDet: A Lightweight SAR Target Detection Algorithm Based on Hybrid Representation Learning Enhancement

In recent years, deep learning has been widely used in remote sensing, especially in the field of synthetic aperture radar (SAR) image target detection. However, all of these deep learning models continue increasing the network’s depth and width without maintaining a good balance between accuracy and speed. Therefore, in this paper, we propose a hybrid representation learning-enhanced SAR target detection algorithm based on the unique features of SAR images from a lightweight perspective called HRLE-SARDet. First, we design a lightweight and scattering feature extraction backbone that is more suitable for SAR image data. Second, for the multiscale feature discrepancy, we design a new multiscale feature fusion neck. Next, to better extract the scattering information from small targets of SAR images and improve the detection accuracy, we design a lightweight hybrid representation learning enhancement module. Finally, to better fit target detection for SAR image datasets, we redesign a more flexible loss function, which allows for an easy adjustment of the importance of polynomial bases according to the target task and dataset. Extensive experimental results on three SAR image ship target datasets (SSDD, AIR-SARShip-2.0, and HRSID) and a newly released large multiclass target SAR dataset (MSAR-1.0) show that our HRLE-SARDet achieves 98.4%, 79.2%, 92.5%, and 88.4% mAPs (mean Average Precision) with only 1.09 M parameters and 2.5 G floating-point operations (FLOPs) on the SSDD, AIR-SARShip-2.0, HRSID, and MSAR-1.0 datasets, which is an excellent performance.

Structural projection points estimation and context priors for oil tank storage estimation in SAR image

3D information extraction is a key direction in the development of Synthetic Aperture Radar (SAR) technology. Extracting the implicit 3D information in SAR image, such as the structure, texture, and occlusion relationship between the environment and the targets, is of significant and challenging for SAR image target interpretation, refined structure parameter extraction and 3D reconstruction. In this article, with a focus on the storage estimation of oil tank in SAR image, we modeled the structural projection points description operator to estimate the fine 3D structural parameters of oil tank and invert the storage information. We proposed the SAR image context prior to further extract the 3D information of the oil tank targets. In addition, by improving the separate offset regressor and adding different attention mechanisms in different branches, we better improved the estimation precision of the structural projection point of the oil tank target. We evaluated our model on the SAR image datasets from RADARSAT-2 with different observation conditions and resolutions, which demonstrates that our proposed model achieves the AP of 87.2% compared with 82.8% of baseline. Finally, we further discussed the effectiveness of the proposed method, as well as the future application prospect and promoting role in the field of SAR image 3D information extraction.

Filtered Convolution for Synthetic Aperture Radar Images Ship Detection

Synthetic aperture radar (SAR) image ship detection is currently a research hotspot in the field of national defense science and technology. However, SAR images contain a large amount of coherent speckle noise, which poses significant challenges in the task of ship detection. To address this issue, we propose filter convolution, a novel design that replaces the traditional convolution layer and suppresses coherent speckle noise while extracting features. Specifically, the convolution kernel of the filter convolution comes from the input and is generated by two modules: the kernel-generation module and local weight generation module. The kernel-generation module is a dynamic structure that generates dynamic convolution kernels using input image or feature information. The local weight generation module is based on the statistical characteristics of the input images or features and is used to generate local weights. The introduction of local weights allows the extracted features to contain more local characteristic information, which is conducive to ship detection in SAR images. In addition, we proved that the fusion of the proposed kernel-generation module and the local weight module can suppress coherent speckle noise in the SAR image. The experimental results show the excellent performance of our method on a large-scale SAR ship detection dataset-v1.0 (LS-SSDD-v1.0). It also achieved state-of-the-art performance on a high-resolution SAR image dataset (HRSID), which confirmed its applicability.

Calving fronts and where to find them: a benchmark dataset and methodology for automatic glacier calving front extraction from synthetic aperture radar imagery

Abstract. Exact information on the calving front positions of marine- or lake-terminating glaciers is a fundamental glacier variable for analyzing ongoing glacier change processes and assessing other variables like frontal ablation rates. In recent years, researchers started implementing algorithms that can automatically detect the calving fronts on satellite imagery. Most studies use optical images, as calving fronts are often easy to distinguish in these images due to the sufficient spatial resolution and the presence of different spectral bands, allowing the separation of ice features. However, detecting calving fronts on synthetic aperture radar (SAR) images is highly desirable, as SAR images can also be acquired during the polar night and are independent of weather conditions (e.g., cloud cover), facilitating year-round monitoring worldwide. In this paper, we present a benchmark dataset (Gourmelon et al., 2022b) of SAR images from multiple regions of the globe with corresponding manually defined labels providing information on the position of the calving front (https://doi.org/10.1594/PANGAEA.940950). With this dataset, different approaches for the detection of glacier calving fronts can be implemented, tested, and their performance fairly compared so that the most effective approach can be determined. The dataset consists of 681 samples, making it large enough to train deep learning segmentation models. It is the first dataset to provide long-term glacier calving front information from multi-mission data. As the dataset includes glaciers from Antarctica, Greenland, and Alaska, the wide applicability of models trained and tested on this dataset is ensured. The test set is independent of the training set so that the generalization capabilities of the models can be evaluated. We provide two sets of labels: one binary segmentation label to discern the calving front from the background, and one label for multi-class segmentation of different landscape classes. Unlike other calving front datasets, the presented dataset contains not only the labels but also the corresponding preprocessed and geo-referenced SAR images as PNG files. The ease of access to the dataset will allow scientists from other fields, such as data science, to contribute their expertise. With this benchmark dataset, we enable comparability between different front detection algorithms and improve the reproducibility of front detection studies. Moreover, we present one baseline model for each kind of label type. Both models are based on the U-Net, one of the most popular deep learning segmentation architectures. In the following two post-processing procedures, the segmentation results are converted into 1-pixel-wide front delineations. By providing both types of labels, both approaches can be used to address the problem. To assess the performance of different models, we suggest first reviewing the segmentation results using the recall, precision, F1 score, and the Jaccard index. Second, the front delineation can be evaluated by calculating the mean distance error to the labeled front. The presented vanilla models provide a baseline of 150 m ± 24 m mean distance error for the Mapple Glacier in Antarctica and 840 m ± 84 m for the Columbia Glacier in Alaska, which has a more complex calving front, consisting of multiple sections, compared with a laterally well constrained, single calving front of Mapple Glacier.

Triangle Distance IoU Loss, Attention-Weighted Feature Pyramid Network, and Rotated-SARShip Dataset for Arbitrary-Oriented SAR Ship Detection

In synthetic aperture radar (SAR) images, ship targets are characterized by varying scales, large aspect ratios, dense arrangements, and arbitrary orientations. Current horizontal and rotation detectors fail to accurately recognize and locate ships due to the limitations of loss function, network structure, and training data. To overcome the challenge, we propose a unified framework combining triangle distance IoU loss (TDIoU loss), an attention-weighted feature pyramid network (AW-FPN), and a Rotated-SARShip dataset (RSSD) for arbitrary-oriented SAR ship detection. First, we propose a TDIoU loss as an effective solution to the loss-metric inconsistency and boundary discontinuity in rotated bounding box regression. Unlike recently released approximate rotational IoU losses, we derive a differentiable rotational IoU algorithm to enable back-propagation of the IoU loss layer, and we design a novel penalty term based on triangle distance to generate a more precise bounding box while accelerating convergence. Secondly, considering the shortage of feature fusion networks in connection pathways and fusion methods, AW-FPN combines multiple skip-scale connections and attention-weighted feature fusion (AWF) mechanism, enabling high-quality semantic interactions and soft feature selections between features of different resolutions and scales. Finally, to address the limitations of existing SAR ship datasets, such as insufficient samples, small image sizes, and improper annotations, we construct a challenging RSSD to facilitate research on rotated ship detection in complex SAR scenes. As a plug-and-play scheme, our TDIoU loss and AW-FPN can be easily embedded into existing rotation detectors with stable performance improvements. Experiments show that our approach achieves 89.18% and 95.16% AP on two SAR image datasets, RSSD and SSDD, respectively, and 90.71% AP on the aerial image dataset, HRSC2016, significantly outperforming the state-of-the-art methods.