The datasets collected so far are all about ship detection. If there are other new data sets found, I will add them to this repository. If you find any links are unavailable, please contact me at: lijt55@mail2.sysu.edu.cn
[Paper]: LS-SSDD-v1.0: A Deep Learning Dataset Dedicated to Small Ship Detection from Large-Scale Sentinel-1 SAR Images
This is a Large-Scale SAR Ship Detection Dataset-v1.0 (LS-SSDD-v1.0) from Sentinel-1, for small ship detection under large-scale backgrounds. LS-SSDD-v1.0 contains 15 large-scale SAR images whose ground truths are correctly labeled by SAR experts by drawing support from Automatic Identification System (AIS) and Google Earth. To facilitate network training, the large-scale images are directly cut into 9,000 sub-images without bells and whistles, providing convenience for subsequent detection result presentation in large-scale SAR images.
LS-SSDD-v1.0 is publicly available at:
Google Cloud: https://drive.google.com/file/d/1-fSKYsKr5wuYuXJE2CDX_Yfr4Ylwjfqm/view?usp=sharing
Baidu Cloud: https://pan.baidu.com/s/1VzqyQHIZL1uPaHmSnRLxPg (Extraction code: 2020)
@article{DBLP:journals/remotesensing/ZhangZKZSWPLSZK20,
author = {Tianwen Zhang and
Xiaoling Zhang and
Xiao Ke and
Xu Zhan and
Jun Shi and
Shunjun Wei and
Dece Pan and
Jianwei Li and
Hao Su and
Yue Zhou and
Durga Kumar},
title = {LS-SSDD-v1.0: {A} Deep Learning Dataset Dedicated to Small Ship Detection
from Large-Scale Sentinel-1 {SAR} Images},
journal = {Remote. Sens.},
volume = {12},
number = {18},
pages = {2997},
year = {2020},
url = {https://doi.org/10.3390/rs12182997},
doi = {10.3390/rs12182997},
bibsource = {dblp computer science bibliography, https://dblp.org}
}[Paper]: A SAR Dataset of Ship Detection for Deep Learning under Complex Backgrounds
This dataset labeled by SAR experts was created using 102 Chinese Gaofen-3 images and 108 Sentinel-1 images. It consists of 43,819 ship chips of 256 pixels in both range and azimuth. These ships mainly have distinct scales and backgrounds. It can be used to develop object detectors for multi-scale and small object detection. SAR-Ship-Dataset is publicly available at: Baidu Cloud (from issues): https://pan.baidu.com/s/1PhSMkXVcuRM8M8xL15iBIQ
@article{DBLP:journals/remotesensing/WangWZDW19a,
author = {Yuanyuan Wang and
Chao Wang and
Hong Zhang and
Yingbo Dong and
Sisi Wei},
title = {A {SAR} Dataset of Ship Detection for Deep Learning under Complex
Backgrounds},
journal = {Remote. Sens.},
volume = {11},
number = {7},
pages = {765},
year = {2019},
url = {https://doi.org/10.3390/rs11070765},
doi = {10.3390/rs11070765},
}Part of SAR-Ship-Dataset
[Paper]: A SAR Dataset of Ship Detection for Deep Learning under Complex Backgrounds
[Paper]: Ship detection in SAR images based on an improved faster R-CNN
Baidu Cloud: https://pan.baidu.com/s/1E8ixqK5AVfXc98UgQmpqaw (Extraction code: trnt, Password: 12345qwert)
[Paper]: Ship detection in SAR images based on an improved faster R-CNN
Baidu Cloud: https://pan.baidu.com/s/1jxOjbRqmmJfQnHHTieHSpA (Extraction code: f7zj)
Here only the annotations file is provided, and the images are the same with SSDD Dataset.
[Paper]: HRSID: A High-Resolution SAR Images Dataset for Ship Detection and Instance Segmentation
High resolution sar images dataset (HRSID) is a data set for ship detection, semantic segmentation, and instance segmentation tasks in high-resolution SAR images. This dataset contains a total of 5604 high-resolution SAR images and 16951 ship instances. HRSID dataset draws on the construction process of the Microsoft Common Objects in Context (COCO) datasets, including SAR images with different resolutions, polarizations, sea conditions, sea areas, and coastal ports. This dataset is a benchmark for researchers to evaluate their approaches. For HRSID, the resolution of SAR images is as follows: 0.5m, 1 m, and 3 m.
HRSID dataset(JPG) can be downloaded from: Baidu cloud: https://pan.baidu.com/s/1z0-E3Lwdkg14feu1FwRf3g (Extraction code: bs64) Google cloud: https://drive.google.com/open?id=1BZTU8Gyg20wqHXtBPFzRazn_lEdvhsbE
HRSID can be downloaded from Google Cloud: Compared to the latest released version of jpg format images of HRSID, we have added the annotations for inshore and offshore ships. Statistically, inshore scenes occupy the proportion of 18.4%, and offshore scenes occupy the proportion of 81.6%. It can be downloaded from URL https://drive.google.com/file/d/1NY3ovgc-woDlNoQdyqzRB3t9McOBH5Ms/view?usp=sharing
As full negative SAR images can provide the information of land or sea clutter, we have prepared 400 SAR images with pure background for testing the robustness of the trained model. It can be downloaded from the URL https://drive.google.com/file/d/1U0Sj1SHoq-2VjXXUKwpXae6rBI3YjyDP/view?usp=sharing
Here is the other version HRSID with png format images. It takes up much more storage space than the version of HRSID with jpg format images, while the high-fidelity SAR images are stored here. This version can be downloaded from the URL: https://drive.google.com/file/d/1xgXi8KC3MDWuu7Yp4n2J-LOPYLRszHAc/view?usp=sharing
HRSID can be downloaded from Baidu Cloud: The files can also be downloaded from URL:Baidu cloud. Extraction code: vxrs https://pan.baidu.com/s/1EitHYJ7tpGh0q9Qb3QA5SQ
[Paper]: AIR-SARShip-1.0：高分辨率SAR舰船检测数据集
[Paper]: AIR-SARShip-1.0: High Resolution SAR Ship Detection Dataset
高分辨率SAR舰船检测数据集-1.0（AIR-SARShip-1.0）首批发布31幅图像，图像分辨率包括1m和3m，成像模式包括聚束式和条带式，极化方式为单极化，场景类型包含港口、岛礁、不同等级海况的海面，目标覆盖运输船、油船、渔船等十余类近千艘舰船。
图像尺寸约为3000×3000像素，图像格式为Tiff、单通道、8/16位图像深度，标注文件提供相应图像的长宽尺寸、标注目标的类别以及标注矩形框的位置。
@articleInfo{R19097,
title = "AIR-SARShip-1.0: High-resolution SAR Ship Detection Dataset (in English)",
journal = "Journal of Radars",
volume = "8",
number = "R19097,
pages = "852",
year = "2019",
note = "",
issn = "2095-283X",
doi = "10.12000/JR19097",
url = "http://radars.ie.ac.cn/en/article/doi/10.12000/JR19097",
author = "SUN Xian","WANG Zhirui","SUN Yuanrui","DIAO Wenhui","ZHANG Yue","FU Kun",keywords = "SAR ship detection","Public dataset","Deep learning",[Paper]: AIR-SARShip-1.0：高分辨率SAR舰船检测数据集
[Paper]: AIR-SARShip-1.0: High Resolution SAR Ship Detection Dataset
高分辨率SAR舰船检测数据集-2.0（AIR-SARShip-2.0）发布300幅图像，图像分辨率包括1m和3m，成像模式包括聚束式和条带式，极化方式为单极化，极化方式为VV，场景类型包含港口、岛礁、不同等级海况的海面，目标覆盖运输船、油船、渔船等十余类数千艘舰船。
图像尺寸约为1000×1000像素，图像格式为Tiff、单通道、8/16位图像深度，标注文件提供相应图像的长宽尺寸、标注目标的类别以及标注矩形框的位置。
@articleInfo{R19097,
title = "AIR-SARShip-1.0: High-resolution SAR Ship Detection Dataset (in English)",
journal = "Journal of Radars",
volume = "8",
number = "R19097,
pages = "852",
year = "2019",
note = "",
issn = "2095-283X",
doi = "10.12000/JR19097",
url = "http://radars.ie.ac.cn/en/article/doi/10.12000/JR19097",
author = "SUN Xian","WANG Zhirui","SUN Yuanrui","DIAO Wenhui","ZHANG Yue","FU Kun",keywords = "SAR ship detection","Public dataset","Deep learning",[Paper]: Very deep learning for ship discrimination in Synthetic Aperture Radar imagery
This dataset was created to help facilitate meaningful comparisons between SAR ship detection and discrimination methods but also provides additional information about some of the ships within the dataset from matched AIS transmissions. These matched ships allow for additional research into information extraction from SAR medium resolution imagery. This distribution of the dataset consists of only the ships and not the raw SAR imagery.
[Paper]: FUSAR-Ship: building a high-resolution SAR-AIS matchup dataset of Gaofen-3 for ship detection and recognition
The FUSAR-Ship high-resolution GF-3 SAR dataset is constructed by running the procedure on a total of 126 GF-3 scenes covering a large variety of sea, land, coast, river and island scenarios. It includes more than 5000 ship chips with AIS messages as well as samples of strong scatterer, bridge, coastal land, islands, sea and land clutter. FUSAR-Ship is intended as an open benchmark dataset for ship and marine target detection and recognition.