Article
Engineering, Marine
Dan Luo, Peng Chen, Jingsong Yang, Xiunan Li, Yizhi Zhao
Summary: This paper presents a supervised learning-based method for ship trajectory classification. A new ensemble classifier is proposed using a machine learning algorithm, and its performance is evaluated using ten-fold cross validation. The results show that the classifier has good performance in ship trajectory classification.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Economics
Huanhuan Li, Hang Jiao, Zaili Yang
Summary: This paper systematically analyzes the performance of ship trajectory prediction methods and conducts experimental tests to compare their prediction performance in real-world scenarios. The results provide a novel perspective and benchmark for ship trajectory prediction research, contributing to maritime safety and autonomous shipping development.
TRANSPORTATION RESEARCH PART E-LOGISTICS AND TRANSPORTATION REVIEW
(2023)
Article
Remote Sensing
Weimin Kong, Shanwei Liu, Mingming Xu, Muhammad Yasir, Dawei Wang, Wantao Liu
Summary: Ship detection in SAR images is important in both military and civil applications. This study proposes a lightweight ship detection network based on the YOLOx-Tiny model to address the challenges of complex backgrounds, varying ship sizes, and real-time detection. The proposed methods include a multi-scale ship feature extraction module that effectively improves the detection accuracy and a SAR remote sensing image detection strategy based on adaptive threshold to suppress false alarms and improve detection speed. Experimental results demonstrate the effectiveness and superiority of the proposed method, providing theoretical and technical support for ship detection in resource-limited platforms and showing promising application prospects.
INTERNATIONAL JOURNAL OF REMOTE SENSING
(2023)
Article
Environmental Sciences
Zhenguo Yan, Xin Song, Lei Yang, Yitao Wang
Summary: The study proposes a SAR image ship classification method based on multiple classifiers ensemble learning (MCEL) and AIS data transfer learning, which increases the ship classification accuracy with limited samples by training models on AIS data and transferring them to SAR images.
Review
Environmental Sciences
Tianwen Zhang, Xiaoling Zhang, Jianwei Li, Xiaowo Xu, Baoyou Wang, Xu Zhan, Yanqin Xu, Xiao Ke, Tianjiao Zeng, Hao Su, Israr Ahmad, Dece Pan, Chang Liu, Yue Zhou, Jun Shi, Shunjun Wei
Summary: SSDD is the first open dataset widely used for research on ship detection from SAR imagery based on deep learning, with 46.59% of public reports confidently choosing it. However, the initial version's coarse annotations and ambiguous usage standards hinder fair methodological comparisons and effective academic exchanges. To address these challenges, SSDD will be officially released in three versions to cater to different research needs.
Article
Telecommunications
Vishal Gupta, Monish Gupta, Parveen Singla
Summary: The study introduces a robust method for ship classification using a combination of Support Vector Machine, bag of features, and Convolutional Neural Networks, achieving a high accuracy of 91.04%. By integrating different algorithms and techniques, the system is able to effectively distinguish ships from cluttered images.
WIRELESS PERSONAL COMMUNICATIONS
(2021)
Article
Chemistry, Multidisciplinary
Yan Shi, Cheng Long, Xuexi Yang, Min Deng
Summary: With the development of navigation globalization and ship dehumanization, there is an increasing demand for ship behavior supervision. This study focuses on detecting abnormal ship behavior based on moving ship trajectory data using spatial and thematic information. A framework is proposed and effectively detects abnormal ship behavior.
APPLIED SCIENCES-BASEL
(2022)
Article
Environmental Sciences
Xiaomeng Geng, Lingli Zhao, Lei Shi, Jie Yang, Pingxiang Li, Weidong Sun
Summary: Ship detection in nearshore areas faces challenges such as material scattering, data labeling difficulties, and interference. A novel method based on candidate target detection, boundary box optimization, and active learning strategy is proposed to achieve high accuracy and efficiency in complex nearshore regions.
Article
Environmental Sciences
Solomiia Kurchaba, Jasper van Vliet, Fons J. Verbeek, Cor J. Veenman
Summary: This study presents a methodology for the automated selection of potentially non-compliant ships using machine learning models on TROPOMI satellite data. The approach involves predicting the amount of NO2 produced by a ship using a regression model and comparing it with actual observations to determine the worthiness of inspection. The results are compared with a segmentation-based method to further identify ships that require attention.
REMOTE SENSING OF ENVIRONMENT
(2023)
Article
Geochemistry & Geophysics
Tamara Loran, Andre Barros Cardoso da Silva, Sushil Kumar Joshi, Stefan V. Baumgartner, Gerhard Krieger
Summary: Near real-time ship monitoring is essential for safety and security at sea. This letter presents two object-oriented ship detectors based on the faster region-based convolutional neural network (R-CNN), which can improve ship detection in radar imagery. The detectors are trained on real X-band airborne RC radar data patches and show high recall performance even in dense multitarget scenarios.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
David Sanchez Pedroche, Jesus Garcia Herrero, Jose Manuel Molina Lopez
Summary: This paper presents a context information extraction process based on actual ship data from AIS, extracting representative points of each trajectory using trajectory segmentation algorithms and obtaining a series of centroids using the k-means algorithm. These centroids form a new representative trajectory that can extract new contextual information from the original set of trajectories, allowing the application of anomaly detection approaches.
Article
Geochemistry & Geophysics
Ming Zhao, Xin Zhang, Andre Kaup
Summary: This article proposes a multitask learning framework for ship detection in SAR images, which includes object detection, speckle suppression, and target segmentation tasks. Various techniques, such as angle classification loss, dual-feature fusion attention, and rotated Gaussian mask, are utilized to improve the accuracy, robustness, and efficiency of ship detection in SAR images. Extensive experiments on SAR ship detection datasets demonstrate the effectiveness and superiority of the proposed method.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Xiangguang Leng, Kefeng Ji, Gangyao Kuang
Summary: In this article, a ship detection method for raw SAR echo data is proposed based on a nonimaging target sensing paradigm. By analyzing the one-dimensional sequence data, ships can be detected without the need for SAR imaging, opening up new possibilities for ship detection in the ocean.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Haoyuan Guo, Xi Yang, Nannan Wang, Xinbo Gao
Summary: This study addresses the challenges of ship detection in SAR images by proposing an effective and stable single-stage detector called CenterNet++, which utilizes modules like feature refinement, feature pyramids fusion, and head enhancement to tackle issues such as small object detection, generating powerful semantic information, and achieving a balance between foreground and background. The proposed method demonstrates state-of-the-art performance on popular SAR image datasets with remarkable accuracy improvement and minimal increase in time cost compared to baseline approaches.
PATTERN RECOGNITION
(2021)
Review
Automation & Control Systems
Huanhuan Li, Hang Jiao, Zaili Yang
Summary: This paper conducts a comparative analysis of the latest ship trajectory prediction algorithms based on machine learning and deep learning methods. It provides valuable guidance and insights for the selection of appropriate prediction methods and discusses the research difficulties and solutions for ship trajectory prediction.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Environmental Sciences
Seung Hee Kim, Hyun-Cheol Kim, Chang-Uk Hyun, Sungjae Lee, Jung-Seok Ha, Joo-Hong Kim, Young-Joo Kwon, Jeong-Won Park, Hyangsun Han, Seong-Yeob Jeong, Duk-jin Kim
Article
Chemistry, Analytical
Ji-hwan Hwang, Duk-jin Kim
Article
Environmental Sciences
Suresh Krishnan Palanisamy Vadivel, Duk-jin Kim, Jungkyo Jung, Yang-Ki Cho, Ki-Jong Han
Summary: The study utilized InSAR technique to assess the VLM at tide gauges in Korea, revealing overall stability with the largest VLM observed at the Pohang tide gauge station. Higher rates of uplift were observed along the coast of the Yellow Sea, while higher rates of subsidence were observed at Jeju and Seogwipo tide gauges. The approach provides unprecedented spatial and temporal resolution for estimating VLM rates at selected tide gauges when in-situ and GNSS observations are not available.
Article
Geosciences, Multidisciplinary
Seung-Woo Lee, Sung Hyun Nam, Duk-Jin Kim
Summary: This study presents a new algorithm for estimating typhoon winds using multiple satellite observations and applies it to Typhoon Soulik (2018). The algorithm showed reasonable and practical estimates in open ocean conditions, but significantly overestimated parameters when the typhoon rapidly weakened before making landfall. The research highlights the importance of continuously monitoring typhoon winds in real-time using multiple satellite observations for timely and operationally important analysis results.
FRONTIERS OF EARTH SCIENCE
(2022)
Review
Environmental Sciences
Do-Seong Byun, Jin-Yong Jeong, Duk-jin Kim, Sungmin Hong, Kyu-Tae Lee, Kitack Lee
Summary: The Ieodo Ocean Research Station provides a platform for monitoring air and sea environments, with technical lessons learned from five research projects launched since 2016. The purpose is to share experiences and best practices to facilitate future research activities in similar environments.
FRONTIERS IN MARINE SCIENCE
(2021)
Article
Engineering, Electrical & Electronic
Juyoung Song, Duk-jin Kim, Sangho An, Junwoo Kim
Summary: This study proposes two novel algorithms to improve the accuracy of vessel detection in synthetic aperture radar (SAR) images. The first algorithm compares the vessel detection output with traditional vessel monitoring apparatus information to demonstrate the position and velocity of vessels. The second algorithm restores the position of the vessel by estimating velocity and measuring the orientation angle. These algorithms show more accurate results compared to traditional methods in the experiments.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Multidisciplinary Sciences
Seung-Tae Lee, Yang-Ki Cho, Duk-jin Kim
Summary: Sea surface temperature (SST) is crucial for understanding coastal seas. Landsat 8 data helps to determine the variability of SST near tidal flats, where the temperature range is higher and the gradients are influenced by heating and cooling from the flats.
SCIENTIFIC REPORTS
(2022)
Article
Environmental Sciences
Junwoo Kim, Hwisong Kim, Hyungyun Jeon, Seung-Hwan Jeong, Juyoung Song, Suresh Krishnan Palanisamy Vadivel, Duk-jin Kim
Summary: The research presents a novel deep learning-based water body extraction model that utilizes Sentinel-1 data and various flood-related geospatial data, showing improved accuracy of up to 7.68% when compared to traditional methods. By customizing and optimizing the U-Net model to incorporate geospatial data, the study demonstrates the potential for operational flood monitoring using deep learning techniques.
Article
Chemistry, Analytical
Ji-Hwan Hwang, Duk-jin Kim, Ki-Mook Kang
Summary: This paper describes a multifunctional scatterometer system and optimized radar signal processing method for simultaneous observation of various physical oceanographic parameters. By integrating separate measurement functions into a single observation system, the efficiency of system operation and cross-analysis of observation data are improved. The operability of the proposed system was examined through field campaigns, and the observation data was cross-analyzed with in-situ data, showing high accuracy.
Article
Environmental Sciences
Junwoo Kim, Hwisong Kim, Duk-jin Kim, Juyoung Song, Chenglei Li
Summary: This article presents a deep learning-based flood area extraction model for a fully automated flood monitoring system. The model was tested and optimized to improve image segmentation accuracy and reduce processing time. The results demonstrate the operation and robustness of the system in accurately extracting flooded areas and reducing misclassification of constructed facilities and mountain shadows. This research could serve as a valuable reference and benchmark for other countries seeking to build cloud-based flood monitoring systems using deep learning.
Article
Engineering, Electrical & Electronic
Hyoseong Lee, Duk-jin Kim
Summary: Tidal flats are internationally protected areas, but also common sites of accidents. Understanding the geomorphologic characteristics of tidal flats is crucial for visitor safety. This article proposes a practical method to correct distorted digital elevation models (DEMs) using a globally collected DEM, effectively estimating changes in tidal flats over time.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Remote Sensing
Juyoung Song, Duk-jin Kim, Ji-hwan Hwang, Sangho An, Junwoo Kim
Summary: This study emphasizes the importance of acquiring precise position and velocity information from GNSS-INS sensors for obtaining SAR images through BPA. Multiple operations of Kalman Filter were conducted to assess the effective order of sensor noise calibration. Experimental results showed that different orders of Kalman Filter applied to FMCW-SAR raw data can achieve optimum BPA image restoration.
KOREAN JOURNAL OF REMOTE SENSING
(2021)