Article
Engineering, Electrical & Electronic
Xin Zhang, Chunlei Huo, Nuo Xu, Hangzhi Jiang, Yong Cao, Lei Ni, Chunhong Pan
Summary: This study introduces a multitask learning-based object detector (MTL-Det) to improve ship detection performance in SAR images by modeling the problem as three cooperative tasks and utilizing auxiliary subtasks to enhance feature learning. The approach effectively addresses the challenges posed by speckle noise in SAR images and outperforms traditional single-task-based object detectors.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Jiamei Fu, Xian Sun, Zhirui Wang, Kun Fu
Summary: A novel detection method named FBR-Net is proposed in this article, which achieves efficient detection of multiscale SAR ships in complex scenes by eliminating the anchor effect, balancing multiple features, and refining object features.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Nuo Xu, Chunlei Huo, Xin Zhang, Yong Cao, Chunhong Pan
Summary: In this paper, a method is proposed to dynamically learn hyperparameter configurations using deep reinforcement learning for ship detection in SAR images. Experimental results demonstrate the effectiveness and advantage of the proposed approach on two SAR image datasets.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Yu Zhang, Xueqian Wang, Zhizhuo Jiang, Gang Li, You He
Summary: This article proposes a lightweight center-based detector for multiscale ship detection in SAR images. The detector utilizes a multilevel auxiliary supervision (MLAS) structure and combines semantic features with different levels. Experimental results demonstrate that the proposed detector achieves comparable detection accuracy while significantly reducing the computation burden.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Engineering, Electrical & Electronic
Tingxuan Yue, Yanmei Zhang, Pengyun Liu, Yanbing Xu, Chengcheng Yu
Summary: In this article, a two-stage ship detection network is proposed, which can generate high-quality anchors and insert a receptive field enhancement module into the feature pyramid network to improve the detection capability for small ships.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Engineering, Electrical & Electronic
Lin Bai, Cheng Yao, Zhen Ye, Dongling Xue, Xiangyuan Lin, Meng Hui
Summary: This article proposes a novel SAR ship detection network called FEPS-Net, which aims to solve the challenges in ship detection in optical remote sensing images. The network utilizes a feature enhancement pyramid and a shallow feature reconstruction module to enhance weak signals and detect small ships. Experimental results demonstrate the advantages of FEPS-Net in SAR ship detection.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Keunhoon Choi, Taeyong Song, Sunok Kim, Hyunsung Jang, Namkoo Ha, Kwanghoon Sohn
Summary: This letter proposes a deep cascade framework for noise-robust SAR ship detection, which sequentially performs despeckle and detection to improve detection performance. Experimental results validate the effectiveness of the proposed method.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Congxia Zhao, Xiongjun Fu, Jian Dong, Rui Qin, Jiayun Chang, Ping Lang
Summary: This article proposes a lightweight end-to-end network based on morphological network for SAR ship detection. The method suppresses speckle noise and enhances edges using the morphological network, and extracts multi-scale features using the feature pyramid fusion structure, thereby improving the accuracy of ship detection.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Engineering, Electrical & Electronic
Shuai Liu, Pengfei Chen, Yudong Zhang
Summary: Synthetic aperture radar (SAR) ship detection is widely used in various applications, but existing lightweight algorithms have problems such as misjudgment of targets mixed with the background and poor detection performance for targets with few samples. To address these issues, this paper proposes a detection network based on a multiscale feature pyramid network (FPN) that enhances target features, suppresses interference from the background, and reduces misjudgment. Experimental results show that the proposed network outperforms existing algorithms on public datasets.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Pengfei Guo, Turgay Celik, Nanqing Liu, Heng-Chao Li
Summary: Substantial progress has been made in detecting ships of arbitrary orientation in synthetic aperture radar (SAR) images. However, the mainstream method is still limited by the horizontal bounding box (HBB) boundary, which cannot provide scaling information for the length and width of the oriented bounding box (OBB) in an intuitive way. In this study, we propose a novel encode representation to describe the OBB by breaking through the border restriction of the HBB. Our comparative experiments demonstrate that our proposed method achieves superior performance and detection accuracy on two commonly-used benchmark datasets for oriented SAR ship detection dataset named SSDD and high-resolution SAR images dataset (HRSID).
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Environmental Sciences
Xiuqin Li, Dong Li, Hongqing Liu, Jun Wan, Zhanye Chen, Qinghua Liu
Summary: In this paper, a novel deep learning network for SAR ship detection, named attention-guided balanced feature pyramid network (A-BFPN), is proposed to better exploit semantic and multilevel complementary features. Experimental results show that the proposed method is superior to the existing algorithms, especially for multi-scale small ship targets under complex background.
Article
Geochemistry & Geophysics
Biao Hou, Zitong Wu, Bo Ren, Zhonghua Li, Xianpeng Guo, Shuang Wang, Licheng Jiao
Summary: A semisupervised SAR ship detection network, SCLANet, is proposed in this study, which improves the algorithm performance by utilizing unlabeled data. SCLANet trains the unlabeled data through consistency learning and achieves high accuracy in ship detection.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Man Xiao, Zhi He, Xinyuan Li, Anjun Lou
Summary: This article proposes an anchor-free method called Pow-FAN to solve ship detection problems in synthetic aperture radar (SAR) images. By utilizing power transformations and feature alignment guided network, Pow-FAN achieves competitive detection results compared to other state-of-the-art methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geochemistry & Geophysics
Qi Hu, Shaohai Hu, Shuaiqi Liu
Summary: In this paper, an anchor-free framework based on a balance attention network is proposed for multiscale ship detection in SAR images. The method utilizes deformable convolution and local attention module to capture local information of ships, and introduces a nonlocal attention module to extract nonlocal features of the SAR image. The feature pyramid network is used to detect ships of different sizes at different scales.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Lin Bai, Cheng Yao, Zhen Ye, Dongling Xue, Xiangyuan Lin, Meng Hui
Summary: In this letter, a novel anchor-free-based detector named FBUA-Net is proposed to address the redundancy of anchor boxes and challenges in ship scales and inshore backgrounds. By adopting a keypoint-based strategy and global context-guided feature balanced pyramid, it achieves state-of-the-art performance on the SSDD and HRSID datasets.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Changjie Cao, Zongjie Cao, Zongyong Cui
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2020)
Article
Geochemistry & Geophysics
Sihang Dang, Zongjie Cao, Zongyong Cui, Yiming Pi, Nengyuan Liu
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2020)
Article
Geochemistry & Geophysics
Changjie Cao, Zongyong Cui, Liying Wang, Jielei Wang, Zongjie Cao, Jianyu Yang
Summary: In this article, a new architecture of automatic target recognition (ATR) model called cost-sensitive awareness-based ATR (CA-ATR) model is proposed to solve the problem of imbalanced data. The method addresses the issue from both the data and algorithm levels, and experimental results demonstrate its superiority.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Zongyong Cui, Yi Qin, Yating Zhong, Zongjie Cao, Haiyi Yang
Summary: This study presents a target detection method based on iterative outliers and recursive saliency depth. By modeling the features of superpixel regions using conditional entropy, effective target detection in SAR images is achieved through iterative anomaly detection and recursion of saliency depth. The proposed method outperforms CFAR and WIE in terms of detection accuracy and practicality.
Article
Environmental Sciences
Changjie Cao, Zongyong Cui, Zongjie Cao, Liying Wang, Jianyu Yang
Summary: This study proposes an integrated approach of counterfactual sample generation and filtering using generative adversarial networks and multiple SVMs to improve the recognition rate of SAR ATR models with small sample sets. The method dynamically enhances the performance of the recognition model by continuously generating counterfactual target samples while filtering those beneficial to the ATR model. Experimental results demonstrate the effectiveness and advantages of the proposed approach in achieving a recognition performance of 91.27% with a significantly reduced training set size.
Article
Computer Science, Artificial Intelligence
Jielei Wang, Zongyong Cui, Ting Jiang, Changjie Cao, Zongjie Cao
Summary: This paper proposes a lightweight network model for ship target detection in synthetic aperture radar (SAR) imagery. The network structure optimization algorithm based on the multi-objective firefly algorithm (NOFA) is designed to encode the filters of a well-performing ship target detection network into a list of probabilities. The multi-objective firefly optimization algorithm (MFA) further optimizes the probability list to output a set of lightweight network encodings.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Environmental Sciences
Zheng Zhou, Zongyong Cui, Zhipeng Zang, Xiangjie Meng, Zongjie Cao, Jianyu Yang
Summary: This paper proposes an ultra-high precision deep learning network (UltraHi-PrNet) for multi-scale target detection in synthetic aperture radar (SAR) images. The network utilizes scale transfer and scale expansion layers to extract features of targets with different scales, and employs Faster R-CNN for target classification and regression. Experimental results demonstrate that the proposed method achieves better performance in detecting different types of targets and outperforms other methods in terms of accuracy.
Article
Environmental Sciences
Ronghui Zhan, Zongyong Cui
Summary: This paper proposes a ship recognition method based on a deep network for detecting and classifying ship targets in SAR scene images. By introducing a squeeze-and-excitation module and constructing a central focal loss function, the proposed method improves accuracy by addressing the issues of high similarity and class imbalance.
Article
Environmental Sciences
Jielei Wang, Zongyong Cui, Zhipeng Zang, Xiangjie Meng, Zongjie Cao
Summary: In this paper, a network pruning method called absorption pruning is proposed to compress the remote sensing object detection network. Unlike existing methods, absorption pruning only needs to be executed once and selects filters that are easy to learn. Furthermore, a method for pruning ratio adjustment based on object characteristics in remote sensing images is designed to better compress deep neural networks. Experimental results demonstrate the effectiveness of the proposed method.
Article
Geochemistry & Geophysics
Bin Li, Zongyong Cui, Yuxuan Sun, Jianyu Yang, Zongjie Cao
Summary: The traditional SAR/ATR algorithm can classify known class samples in the test set, but catastrophic forgetting can occur when training only with new class samples. This article proposes DCBES, a method for selecting key samples of the old class based on metric learning and set covering theory. Experimental results on the MSTAR dataset show that DCBES outperforms other exemplar selection methods and achieves the best results.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Sihang Dang, Zongyong Cui, Zongjie Cao, Yiming Pi, Xiaoyi Feng
Summary: In order to improve the ATR system effectively when new unknown samples are constantly captured, it is necessary to examine the existing training samples and recognition model so that the system could autonomously assess new unknown samples with low predictive reliability during the recognition process and learn them preferentially.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Bin Li, Zongyong Cui, Zongjie Cao, Jianyu Yang
Summary: This article proposes an incremental class anchor clustering (ICAC) method to address the issue of catastrophic forgetting in SAR ATR. ICAC learns new classes, enables the model to recognize and classify old classes, and solves the imbalance between old and new classes. The method incorporates knowledge distillation and separable learning strategy to improve accuracy.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Changjie Cao, Zongyong Cui, Liying Wang, Jielei Wang, Zongjie Cao, Jianyu Yang
Summary: In this study, a solution method for the problem of imbalanced data in the automatic target recognition (ATR) model is proposed, called demand-driven generative adversarial nets (DDGANs). This method alleviates the negative impact of data imbalance by generating new samples and can autonomously learn the generation demands. It achieves imbalanced data learning for different categories of target samples.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Zongyong Cui, Xiaoya Wang, Nengyuan Liu, Zongjie Cao, Jianyu Yang
Summary: A ship detection method in large-scale SAR images via CenterNet is proposed, which defines the target as a point and locates the center point of the target through key point estimation to avoid missing small targets, and reduces false alarms through the introduction of SSE attention module. Experimental results show that the proposed method can detect all targets in dense-docking conditions.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Liying Wang, Zongjie Cao, Zongyong Cui, Changjie Cao, Yiming Pi
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2020)