RSDD-SAR：SAR舰船斜框检测数据集

徐从安; 苏航; 李健伟; 刘瑜; 姚力波; 高龙; 闫文君; 汪韬阳

doi:10.12000/JR22007

RSDD-SAR：SAR舰船斜框检测数据集

doi: 10.12000/JR22007

徐从安^{1, 2},
苏航^1, ,,
李健伟^3, ,,
刘瑜¹,
姚力波¹,
高龙¹,
闫文君¹,
汪韬阳⁴

1.
海军航空大学烟台 264000
2.
北京理工大学前沿技术研究院济南 250300
3.
92877部队舟山 316000
4.
武汉大学武汉 430000

基金项目: 国家自然科学基金(61790550, 61790554, 61971432, 62022092)，中国科协青年人才托举工程基金(2020-JCJQ-QT-011)，山东省泰山学者人才工程(tsqn201909156)

详细信息

作者简介:
徐从安(1987-)，男，博士，副教授，主要研究方向为多平台多源预警探测、智能信息处理

苏　航(1998-)，男，海军航空大学在读硕士研究生，主要研究方向为SAR图像舰船目标检测

李健伟(1989-)，男，博士，工程师，主要研究方向为雷达与电子对抗、SAR图像智能处理、计算机视觉

刘　瑜(1987-)，男，博士，教授，主要研究方向为遥感图像处理、多模态数据融合

姚力波(1980-)，男，博士，副教授，主要研究方向为卫星遥感图像处理、目标检测与跟踪

高　龙(1993-)，男，博士，讲师，主要研究方向机器学习、异常检测、SAR图像智能处理

闫文君(1986-)，男，博士，副教授，主要研究方向为智能信息处理、频谱感知

汪韬阳(1984-)，男，博士，副研究员，主要研究方向为航天摄影测量、遥感图像几何处理、卫星视频目标检测与识别

通讯作者:
苏航 shpersonal_email@163.com

李健伟 lgm_jw@163.com

责任主编：孙显 Corresponding Editor: SUN Xian
中图分类号: TN957.51; TN958
计量
- 文章访问数: 4340
- HTML全文浏览量: 3396
- PDF下载量: 284
- 被引次数: 0
出版历程
- 收稿日期: 2022-01-09
- 修回日期: 2022-05-17
- 网络出版日期: 2022-06-08
- 刊出日期: 2022-08-28

RSDD-SAR: Rotated Ship Detection Dataset in SAR Images

XU Congan^{1, 2},
SU Hang^{1
, ,},
LI Jianwei^{3
, ,},
LIU Yu¹,
YAO Libo¹,
GAO Long¹,
YAN Wenjun¹,
WANG Taoyang⁴

1.
Naval Aviation University, Yantai 264000, China
2.
Advanced Technology Research Institute, Beijing Institute of Technology, Jinan 250300, China
3.
Troops of 92877, Zhoushan 316000, China
4.
Wuhan University, Wuhan 430000, China

Funds: The National Natural Science Foundation of China (61790550, 61790554, 61971432, 62022092), The Young Elite Scientists Sponsorship Program by CAST (2020-JCJQ-QT-011), The Taishan Scholar Project of Shandong Province (tsqn201909156)

More Information

Corresponding author: SU Hang, shpersonal_email@163.com; LI Jianwei, lgm_jw@163.com

摘要

摘要: 针对合成孔径雷达(SAR)舰船斜框检测数据集较少，难以满足算法发展和实际应用需求的问题，该文公开了SAR舰船斜框检测数据集(RSDD-SAR)，该数据集由84景高分3号数据和41景TerraSAR-X数据切片及2景未剪裁大图，共127景数据构成，包含多种成像模式、多种极化方式、多种分辨率切片7000张，舰船实例10263个，通过自动标注和人工修正相结合的方式高效标注。同时，该文对几种常用的光学遥感图像斜框检测算法和SAR舰船斜框检测算法进行了实验，其中单阶段算法S2ANet检测效果最佳，平均精度达到90.06%。通过实验对比分析形成基准指标，可供相关学者参考。最后，该文通过泛化能力测试，分析讨论了RSDD-SAR数据集训练模型在其他数据集和未剪裁大图上的性能，结果表明：该数据集训练模型具有较好的泛化能力，说明该数据集具有较强的应用价值。RSDD-SAR数据集可在以下网址下载：https://radars.ac.cn/web/data/getData?dataType=SDD-SAR。
- 合成孔径雷达 /
- 舰船斜框检测 /
- 公开数据集 /
- SAR舰船斜框检测数据集 /
- 深度学习
Abstract: This paper releases a rotated SAR ship detection dataset, named Rotated Ship Detection Dataset in SAR Images (RSDD-SAR), to address the problem that the existing rotated SAR ship detection datasets are not enough to meet the requirements of algorithm development and practical application. This dataset consists of 84 scenes of GF-3 data slices, 41 scenes of TerraSAR-X data slices, and 2 scenes of large uncropped images, including 7,000 slices and 10,263 ship instances of multi-observing modes, multi-polarization modes, and multi-resolutions. This dataset is effectively annotated by automatic annotation with manual correction. Meanwhile, experiments were conducted for several popular rotated object detection algorithms in optical remote sensing images and rotated ship detection algorithms in SAR images, and the one-stage algorithm S2ANet achieved the highest average precision of 90.06%. When using this dataset, scholars can reference the experimental results, and corresponding analysis can be used. Finally, this paper conducts generalization ability testing experiments on other datasets and large uncropped images to analyze and discuss the performance of the model trained on RSDD-SAR. The experimental results show that the model trained on RSDD-SAR has decent performance and confirms the application value of this dataset. The RSDD-SAR dataset is available at https://radars.ac.cn/web/data/getData?dataType=SDD-SAR.
- Synthetic Aperture Radar (SAR) /
- Rotated SAR ship detection /
- Public dataset /
- RSDD-SAR /
- Deep learning

HTML全文

图 1 不同标注方式比较

Figure 1. Comparison of different annotation methods

下载: 全尺寸图片幻灯片

图 2 舰船目标航向和长宽比信息

Figure 2. The course and aspect ratio information of ship target

下载: 全尺寸图片幻灯片

图 3 斜框定义方式

Figure 3. Rotated bounding box definition method

下载: 全尺寸图片幻灯片

图 4 标注流程图

Figure 4. Annotation procedure

下载: 全尺寸图片幻灯片

图 5 数据预处理流程

Figure 5. Data preprocessing procedure

下载: 全尺寸图片幻灯片

图 6 数据切片方式

Figure 6. Data cutting method

下载: 全尺寸图片幻灯片

图 7 人工修正示例

Figure 7. Manual modification examples

下载: 全尺寸图片幻灯片

图 8 Google Earth纠正

Figure 8. Google Earth correction

下载: 全尺寸图片幻灯片

图 9 标注示例

Figure 9. Annotation example

下载: 全尺寸图片幻灯片

图 10 RSDD-SAR数据集结构

Figure 10. Structure of RSDD-SAR dataset

下载: 全尺寸图片幻灯片

图 11 数据集舰船角度和长宽比分布图

Figure 11. Angle and aspect ratio distribution maps of ships in RSDD-SAR dataset

下载: 全尺寸图片幻灯片

图 12 RSDD-SAR数据集典型场景

Figure 12. Typical scenarios in RSDD-SAR

下载: 全尺寸图片幻灯片

图 13 不同算法检测结果((a) 标注；(b) R-Faster R-CNN-ResNet-101检测结果；(c) RoI Transformer-ResNet-101检测结果；(d) Gliding Vertex-ResNet-101检测结果；(e) Oriented R-CNN-ResNet-101检测结果；(f) R-RetinaNet-ResNet-101检测结果；(g) S2ANet-ResNet-101检测结果；(h) R3Det-ResNet-101检测结果；(i) Redet-ReResNet-50检测结果；(j) DRBox-V2检测结果；(k) R-FCOS-ResNet-101检测结果；(l) CAF-ResNet-101检测结果；(m) Polar Encoding-ResNet-101检测结果)

Figure 13. Detection results of different methods ((a) Annotations; (b) Detection results of R-Faster R-CNN-ResNet-101; (c) Detection results of RoI Transformer-ResNet-101; (d) Detection results of Gliding Vertex-ResNet-101; (e) Detection results of Oriented R-CNN-ResNet-101; (f) Detection results of R-RetinaNet-ResNet-101; (g) Detection results of S2ANet-ResNet-101; (h) Detection results of R3Det-ResNet-101; (i) Detection results of Redet-ReResNet-50; (j) Detection results of DRBox-V2; (k) Detection results of R-FCOS-ResNet-101; (l) Detection results of CAF-ResNet-101; (m) Detection results of Polar Encoding-ResNet-101)

下载: 全尺寸图片幻灯片

图 14 SSDD数据集测试结果

Figure 14. Testing results on SSDD

下载: 全尺寸图片幻灯片

图 15 S2ANet AP₅₀曲线

Figure 15. AP₅₀ curves of S2ANet

下载: 全尺寸图片幻灯片

图 16 未剪裁大图测试

Figure 16. Testing on uncropped images

下载: 全尺寸图片幻灯片

1 RSDD-SAR数据集发布地址

1. Release address of RSDD-SAR dataset

下载: 全尺寸图片幻灯片

表 1 现有公开数据集详细信息

Table 1. Detailed information of existing public datasets

数据集	公开时间	数据来源	分辨率(m)	图像尺寸	图像数量	任务
OpenSARShip-1.0 OpenSARShip-2.0	2017	Sentinel-1	2.7×22～ 3.5×22, 20×22	9×9～445×445 1×1～445×445	11346 34528	识别
SSDD	2017	RadarSat-2 TerraSAR-X Sentinel-1	1～15	190～668	1160	垂直边框检测斜框检测语义分割
SAR-Ship-Dataset	2019	Gaofen-3 Sentinel-1	3～25	256×256	43918	垂直边框检测
AIR-SARShip-1.0 AIR-SARShip-2.0	2019	Gaofen-3	1, 3	3000×3000 1000×1000	31 300	垂直边框检测
FUSAR-Ship	2020	Gaofen-3	1.700×1.124～ 1.754×1.124	512×512	5243	识别
HRSID	2021	Sentinel-1 TerraSAR-X	0.5, 1.0, 3.0	800×800	5604	垂直边框检测语义分割
LS-SSDD-v1.0	2021	Sentinel-1	5×20	24000×16000	15	垂直边框检测
SRSDD-v1.0	2021	Gaofen-3	1	1024×1024	666	斜框检测识别
RSDD-SAR	2022	Gaofen-3 TerraSAR-X	2～20	512×512	7000	斜框检测

下载: 导出CSV

表 2 原始数据详细信息

Table 2. Detailed information of the raw data

景号	传感器	经度	纬度	成像时间	成像模式	分辨率(m)	极化方式	产品级别	入射角(°)	成像幅宽(km)	编号
1	GF-3	E121.0	N37.9	20171017	FSII	10	HH,HV	L1A	19~50	100	0-1

下载: 导出CSV

表 3 依据COCO划分标准RSDD-SAR舰船尺寸统计

Table 3. Area statistics of ships in RSDD-SAR according to COCO

目标类型	数量	比例
Small (area < 32²)	8331	81.17%
Medium (32² < area < 96²)	1927	18.78%
Large (area > 96²)	5	0.05%

下载: 导出CSV

表 4 依据文献[13]划分标准RSDD-SAR舰船尺寸统计

Table 4. Area statistics of ships in RSDD-SAR according to Ref. [13]

目标类型	数量	比例
Small (area < 625)	6146	59.88%
Medium (625 ≤ area ≤ 7500)	4109	40.04%
Large (area > 7500)	8	0.08%

下载: 导出CSV

表 5 不同算法实验结果

Table 5. Experimental results of different algorithms

模型		骨干网络	Params (M)	FPS	Inshore AP₅₀(%)	Offshore AP₅₀(%)	AP₅₀(%)
两阶段	R-Faster R-CNN	ResNet-101	60.45	13.28	50.99	91.47	84.10
		ResNet-50	41.41	15.87	48.78	90.93	83.29
		ResNet-18	28.30	22.36	43.18	88.57	80.30
	RoI Transformer	ResNet-101	74.37	8.10	64.40	94.85	89.48
		ResNet-50	55.32	11.93	60.83	94.35	88.39
		ResNet-18	42.21	13.71	56.51	93.30	86.60
	Gliding Vertex	ResNet-101	60.45	13.32	62.24	93.50	88.16
		ResNet-50	41.41	16.82	55.93	91.65	85.55
		ResNet-18	28.29	22.33	51.48	91.49	84.63
	Oriented R-CNN	ResNet-101	60.34	21.90	66.77	90.28	88.85
		ResNet-50	41.35	26.60	65.92	90.21	88.84
		ResNet-18	28.28	34.30	61.82	90.05	87.50
	ReDet	ReResNet-50	31.57	15.90	61.94	90.34	88.40
单阶段	R-RetinaNet	ResNet-101	51.49	15.33	35.75	74.92	67.89
		ResNet-50	32.44	22.31	33.20	74.06	66.66
		ResNet-18	19.38	34.15	30.10	72.74	65.09
	S2ANet	ResNet-101	55.50	15.74	66.43	94.94	90.06
		ResNet-50	36.45	23.46	63.27	93.14	87.91
		ResNet-18	19.85	32.06	59.61	92.50	86.88
	R3Det	ResNet-101	60.80	24.10	57.73	90.09	80.92
		ResNet-50	41.81	29.30	56.87	90.16	80.87
		ResNet-18	25.25	37.60	54.92	89.69	80.44
	DRBox-V2	VGG16	15.91	28.28	57.79	91.28	85.63
无锚框	R-FCOS	ResNet-101	51.21	18.85	56.17	93.79	87.31
		ResNet-50	32.17	31.39	50.02	93.09	85.48
		ResNet-18	19.11	42.01	49.48	92.33	84.78
	CFA	ResNet-101	55.82	25.80	67.35	90.33	89.46
		ResNet-50	36.83	36.60	66.40	90.47	89.31
		ResNet-18	20.27	52.20	67.35	90.27	88.97
	Polar Encoding	ResNet-101	71.83	16.71	62.02	89.99	87.88
		ResNet-50	52.83	17.56	59.69	90.12	87.31
		ResNet-18	13.36	27.48	58.05	89.31	85.28

下载: 导出CSV

表 6 泛化能力测试

Table 6. Generalization ability testing results

模型	训练集	验证集	验证集AP₅₀(%)	测试集512×512 AP₅₀(%)	测试集800×800 AP₅₀(%)
S2ANet-ResNet-101	RSDD-SAR训练集	RSDD-SAR测试集	90.06	57.47(–32.59)	63.04(–27.02)
S2ANet-ResNet-101	SSDD训练集	SSDD测试集	90.52	36.24(–54.28)	47.87(–42.65)
S2ANet-ResNet-50	RSDD-SAR训练集	RSDD-SAR测试集	87.91	56.04(–31.87)	63.15(–24.76)
S2ANet-ResNet-50	SSDD训练集	SSDD测试集	92.73	41.81(–50.92)	51.83(–40.90)
S2ANet-ResNet-18	RSDD-SAR训练集	RSDD-SAR测试集	86.88	55.73(–31.15)	62.04(–24.84)
S2ANet-ResNet-18	SSDD训练集	SSDD测试集	90.30	34.52(–55.78)	48.89(–41.41)

下载: 导出CSV

表 7 未剪裁大图测试结果

Table 7. Results on uncropped images

模型	训练集	验证集	验证集AP₅₀(%)	未剪裁大图AP₅₀(%)
S2ANet-ResNet-101	RSDD-SAR训练集	RSDD-SAR测试集	90.06	65.97(–24.09)
S2ANet-ResNet-101	SSDD训练集	SSDD测试集	90.52	51.04(–39.48)

下载: 导出CSV

1 RSDD-SAR数据集详细信息

1. RSDD-SAR dataset information in detail

景号	传感器	经度	纬度	时间	成像模式	分辨率(m)	极化方式	产品级别	入射角(°)	成像幅宽(km)	编号
1	GF-3	E121.0	N37.9	20171017	FSII	10	HH,HV	L1A	19~50	100	0-1
2	GF-3	E119.3	N37.2	20210809	UFS	3	DH	L1A	20~50	30	2
3	GF-3	E121.3	N37.5	20180901	FSI	5	HH,HV	L1A	19~50	50	3-4
4	GF-3	E120.5	N37.9	20210228	FSI	5	VH,VV	L1A	19~50	50	5-6
5	GF-3	E121.0	N35.8	20210228	FSI	5	VH,VV	L1A	19~50	50	7-8
6	GF-3	E119.1	N38.2	20210619	FSI	5	DV	L1A	19~50	50	9
7	GF-3	E120.5	N37.8	20210526	SS	25	HH,HV	L1A	17~50	130	10-11
8	GF-3	E120.5	N35.6	20210716	SS	25	HH,HV	L1A	17~50	130	12-13
9	GF-3	E122.9	N37.5	20191231	UFS	3	DH	L1A	20~50	30	14
10	GF-3	E119.8	N35.0	20210228	UFS	3	DH	L1A	20~50	30	15
11	GF-3	E119.8	N35.3	20210228	UFS	3	DH	L1A	20~50	30	16
12	GF-3	E120.4	N37.9	20210228	UFS	3	DH	L1A	20~50	30	17
13	GF-3	E120.9	N37.9	20210718	FSII	10	HH,HV	L1A	19~50	100	18-19
14	GF-3	E120.3	N36.0	20210305	FSI	5	VH,VV	L1A	19~50	50	20-21
15	GF-3	E121.6	N37.6	20210802	FSI	5	HH,HV	L1A	19~50	50	22-23
16	GF-3	E122.5	N37.5	20210130	SS	25	VH,VV	L1A	17~50	130	24-25
17	GF-3	E120.3	N35.6	20210504	SS	25	VH,VV	L1A	17~50	130	26-27
18	GF-3	E120.6	N36.2	20210519	SS	25	HH,HV	L1A	17~50	130	28-29
19	GF-3	E121.0	N36.0	20210612	SS	25	VH,VV	L1A	17~50	130	30-31
20	GF-3	E121.5	N38.2	20210612	SS	25	VH,VV	L1A	17~50	130	32-33
21	GF-3	E120.1	N36.0	20210723	SS	25	HH,HV	L1A	17~50	130	34-35
22	GF-3	E120.6	N38.1	20210723	SS	25	HH,HV	L1A	17~50	130	36-37
23	GF-3	E119.8	N35.0	20210130	UFS	3	DH	L1A	20~50	30	38
24	GF-3	E119.8	N35.3	20210130	UFS	3	DH	L1A	20~50	30	39
25	GF-3	E119.9	N35.6	20210130	UFS	3	DH	L1A	20~50	30	40
26	GF-3	E120.4	N37.9	20210130	UFS	3	DH	L1A	20~50	30	41
27	GF-3	E121.6	N37.6	20210427	UFS	3	DH	L1A	20~50	30	42
28	GF-3	E118.8	N38.1	20210521	UFS	3	DH	L1A	20~50	30	43
29	GF-3	E122.1	N37.5	20210101	FSI	5	VH,VV	L1A	19~50	50	44-45
30	GF-3	E121.1	N35.8	20210113	FSI	5	VH,VV	L1A	19~50	50	46-47
31	GF-3	E120.9	N36.1	20210125	FSI	5	VH,VV	L1A	19~50	50	48-49
32	GF-3	E120.3	N38.2	20210204	FSI	5	VH,VV	L1A	19~50	50	50-51
33	GF-3	E120.7	N36.1	20210329	SS	25	VH,VV	L1A	17~50	130	52-53
34	GF-3	E119.6	N37.9	20210417	SS	25	VH,VV	L1A	17~50	130	54-55
35	TerraSAR-X	E056	N27	20080311	SL	2	HH	SSC	20~55	10	56
36	TerraSAR-X	E100	N13	20111209	SM	3	HH	MGD	20~45	30×50	57
37	TerraSAR-X	E013	S08	20190730	SM	3	HH	EEC	20~45	30×50	58
38	TerraSAR-X	W090	N29	20100516	SM	3	HH	EEC	20~45	30×50	59
39	TerraSAR-X	W090	N29	20120829	SM	3	HH	EEC	20~45	30×50	60
40	TerraSAR-X	W090	N29	20170807	SM	3	HH	SSC	20~45	30×50	61
41	TerraSAR-X	E023	N37	20180416	SM	3	HH	SSC	20~45	30×50	62
42	TerraSAR-X	E023	N37	20160830	SM	3	HH	SSC	20~45	30×50	63
43	TerraSAR-X	E023	N37	20170305	SM	3	HH	SSC	20~45	30×50	64
44	TerraSAR-X	E023	N37	20170623	SM	3	HH	SSC	20~45	30×50	65
45	TerraSAR-X	E023	N37	20170828	SM	3	HH	SSC	20~45	30×50	66
46	TerraSAR-X	E023	N37	20171022	SM	3	HH	SSC	20~45	30×50	67
47	TerraSAR-X	E023	N37	20171205	SM	3	HH	SSC	20~45	30×50	68
48	TerraSAR-X	E023	N37	20180209	SM	3	HH	SSC	20~45	30×50	69
49	TerraSAR-X	E023	N37	20180621	SM	3	HH	SSC	20~45	30×50	70
50	TerraSAR-X	E023	N37	20180815	SM	3	HH	SSC	20~45	30×50	71
51	TerraSAR-X	E023	N37	20181020	SM	3	HH	SSC	20~45	30×50	72
52	TerraSAR-X	E023	N37	20190127	SM	3	HH	SSC	20~45	30×50	73
53	TerraSAR-X	E023	N37	20190323	SM	3	HH	SSC	20~45	30×50	74
54	TerraSAR-X	E023	N37	20190619	SM	3	HH	SSC	20~45	30×50	75
55	TerraSAR-X	E023	N37	20190904	SM	3	HH	SSC	20~45	30×50	76
56	TerraSAR-X	E023	N37	20160614	SM	3	HH	SSC	20~45	30×50	77
57	TerraSAR-X	E121	N31	20151016	SM	3	VV	SSC	20~45	30×50	78
58	TerraSAR-X	E121	N31	20160614	SM	3	VV	SSC	20~45	30×50	79
59	TerraSAR-X	E121	N31	20160717	SM	3	VV	SSC	20~45	30×50	80
60	TerraSAR-X	E121	N31	20160819	SM	3	VV	SSC	20~45	30×50	81
61	TerraSAR-X	E121	N31	20151118	SM	3	VV	SSC	20~45	30×50	82
62	TerraSAR-X	E121	N31	20151210	SM	3	VV	SSC	20~45	30×50	83
63	TerraSAR-X	E121	N31	20160101	SM	3	VV	SSC	20~45	30×50	84
64	TerraSAR-X	E121	N31	20160203	SM	3	VV	SSC	20~45	30×50	85
65	TerraSAR-X	E121	N31	20160409	SM	3	VV	SSC	20~45	30×50	86
66	TerraSAR-X	E121	N31	20160512	SM	3	VV	SSC	20~45	30×50	87
67	TerraSAR-X	E119	N37	20151220	SM	3	HH	SSC	20~45	30×50	88
68	TerraSAR-X	E023	N37	20120305	SM	3	HH	SSC	20~45	30×50	89
69	TerraSAR-X	E023	N37	20120601	SM	3	HH	SSC	20~45	30×50	90
70	TerraSAR-X	E023	N37	20120908	SM	3	HH	SSC	20~45	30×50	91
71	TerraSAR-X	E023	N37	20121205	SM	3	HH	SSC	20~45	30×50	92
72	TerraSAR-X	E023	N37	20130314	SM	3	HH	SSC	20~45	30×50	93
73	TerraSAR-X	E023	N37	20151129	SM	3	HH	SSC	20~45	30×50	94
74	TerraSAR-X	E023	N37	20160225	SM	3	HH	SSC	20~45	30×50	95
75	TerraSAR-X	E023	N37	20160420	SM	3	HH	SSC	20~45	30×50	96
76	GF-3	E119.3	N35.0	20200619	FSII	10	HH,HV	L1A	19~50	100	97-98
77	GF-3	E120.1	N38.0	20200619	FSII	10	HH,HV	L1A	19~50	100	99-100
78	GF-3	E118.4	N38.0	20200711	FSII	10	HH,HV	L1A	19~50	100	101-102
79	GF-3	E118.0	N38.1	20200718	FSII	10	HH,HV	L1A	19~50	100	103-104
80	GF-3	E118.8	N38.5	20200730	FSII	10	HH,HV	L1A	19~50	100	105-106
81	GF-3	E119.5	N34.9	20200730	FSII	10	HH,HV	L1A	19~50	100	107-108
82	GF-3	E118.9	N37.9	20200804	FSII	10	HH,HV	L1A	19~50	100	109-110
83	GF-3	E117.9	N38.6	20200809	FSII	10	HH,HV	L1A	19~50	100	111-112
84	GF-3	E118.5	N38.2	20200809	FSII	10	HH,HV	L1A	19~50	100	113-114
85	GF-3	E120.2	N35.0	20201001	FSII	10	VH,VV	L1A	19~50	100	115-116
86	GF-3	E120.5	N36.0	20201001	FSII	10	VH,VV	L1A	19~50	100	117-118
87	GF-3	E121.0	N38.0	20201001	FSII	10	VH,VV	L1A	19~50	100	119-120
88	GF-3	E119.7	N35.4	20200804	FSI	5	HH,HV	L1A	19~50	50	121-122
89	GF-3	E119.8	N35.0	20200804	FSI	5	HH,HV	L1A	19~50	50	123-124
90	GF-3	E120.8	N38.4	20201128	FSI	5	VH,VV	L1A	19~50	50	125-126
91	GF-3	E120.9	N37.9	20201128	FSI	5	VH,VV	L1A	19~50	50	127-128
92	GF-3	E121.3	N36.0	20201128	FSI	5	VH,VV	L1A	19~50	50	129-130
93	GF-3	E118.1	N38.2	20200525	FSI	5	HH,HV	L1A	19~50	50	131-132
94	GF-3	E120.2	N35.7	20200602	SS	25	VH,VV	L1A	17~50	130	133-134
95	GF-3	E119.5	N35.7	20200711	FSII	10	VH,VV	L1A	19~50	100	135-136
96	GF-3	E118.8	N38.3	20200728	FSII	10	HH,HV	L1A	19~50	100	137-138
97	GF-3	E119.6	N35.3	20200728	FSII	10	HH,HV	L1A	19~50	100	139-140
98	GF-3	E120.2	N35.4	20200519	SS	25	VH,VV	L1A	17~50	130	141-142
99	GF-3	E119.9	N37.7	20200521	SS	25	VH,VV	L1A	17~50	130	143-144
100	GF-3	E120.2	N35.8	20200521	SS	25	VH,VV	L1A	17~50	130	145-146
101	GF-3	E120.4	N34.9	20200521	SS	25	VH,VV	L1A	17~50	130	147-148
102	GF-3	E121.0	N35.8	20200502	FSII	10	VH,VV	L1A	19~50	100	149-150
103	GF-3	E120.5	N35.9	20200509	FSII	10	VH,VV	L1A	19~50	100	151-152
104	GF-3	E120.0	N38.5	20200514	FSII	10	VH,VV	L1A	19~50	100	153-154
105	GF-3	E120.0	N38.3	20200729	FSII	10	HH,HV	L1A	19~50	100	155-156
106	GF-3	E120.6	N36.0	20201006	FSII	10	VH,VV	L1A	19~50	100	157-158
107	GF-3	E122.8	N37.2	20200924	QPSI	8	HH,HV,VH,VV	L1A	20~41	30	159-162
108	GF-3	E120.0	N35.4	20201030	QPSI	8	HH,HV,VH,VV	L1A	20~41	30	163-166
109	GF-3	E119.1	N37.4	20200814	UFS	3	DH	L1A	20~50	30	167
110	GF-3	E119.2	N38.0	20200814	UFS	3	DH	L1A	20~50	30	168
111	GF-3	E120.1	N37.8	20200926	UFS	3	DH	L1A	20~50	30	169
112	GF-3	E120.5	N36.0	20200926	UFS	3	DH	L1A	20~50	30	170
113	GF-3	E121.3	N37.7	20201121	UFS	3	DH	L1A	20~50	30	171
114	GF-3	E121.6	N36.6	20201121	UFS	3	DH	L1A	20~50	30	172
115	GF-3	E120.0	N38.5	20201104	FSII	10	VH,VV	L1A	19~50	100	173-174
116	GF-3	E121.1	N35.8	20201126	FSI	5	VH,VV	L1A	19~50	50	175-176
117	GF-3	E121.5	N37.7	20201126	FSI	5	VH,VV	L1A	19~50	50	177-178
118	GF-3	E121.4	N38.1	20200519	SS	25	VH,VV	L1A	17~50	130	179-180
119	GF-3	E120.9	N35.4	20200531	SS	25	VH,VV	L1A	17~50	130	181-182
120	GF-3	E119.8	N37.9	20200619	SS	25	VH,VV	L1A	17~50	130	183-184
121	GF-3	E120.4	N35.1	20200619	SS	25	VH,VV	L1A	17~50	130	185-186
122	GF-3	E120.3	N34.8	20200704	SS	25	VH,VV	L1A	17~50	130	187-188
123	GF-3	E120.5	N35.8	20200704	SS	25	VH,VV	L1A	17~50	130	189-190
124	GF-3	E121.9	N37.7	20201011	UFS	3	DH	L1A	20~50	30	191
125	GF-3	E118.0	N38.3	20201111	UFS	3	DH	L1A	20~50	30	192
126	GF-3	E119.8	N35.2	20210619	FSI	5	DV	L1A	19~50	50	GF3
127	TerraSAR-X	E119	N37	20151128	SM	3	HH	SSC	20~45	30×50	Terra

下载: 导出CSV

参考文献(40)

[1]	LI Jianwei, QU Changwen, and SHAO Jiaqi. Ship detection in SAR images based on an improved faster R-CNN[C]. 2017 SAR in Big Data Era: Models, Methods and Applications (BIGSARDATA), Beijing, China, 2017: 1–6.
[2]	ZHANG Xiaohan, WANG Haipeng, XU Congan, et al. A lightweight feature optimizing network for ship detection in SAR image[J]. IEEE Access, 2019, 7: 141662–141678. doi: 10.1109/ACCESS.2019.2943241
[3]	ZHANG Tianwen, ZHANG Xiaoling, SHI Jun, et al. Depthwise separable convolution neural network for high-speed SAR ship detection[J]. Remote Sensing, 2019, 11(21): 2483. doi: 10.3390/rs11212483
[4]	ZHANG Peng, TANG Jinsong, ZHONG Heping, et al. Self-trained target detection of radar and sonar images using automatic deep learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 4701914. doi: 10.1109/TGRS.2021.3096011
[5]	孙显, 王智睿, 孙元睿, 等. AIR-SARShip-1.0: 高分辨率SAR舰船检测数据集[J]. 雷达学报, 2019, 8(6): 852–862. doi: 10.12000/JR19097 SUN Xian, WANG Zhirui, SUN Yuanrui, et al. AIR-SARShip-1.0: High-resolution SAR ship detection dataset[J]. Journal of Radars, 2019, 8(6): 852–862. doi: 10.12000/JR19097
[6]	WANG Yuanyuan, WANG Chao, ZHANG Hong, et al. A SAR dataset of ship detection for deep learning under complex backgrounds[J]. Remote Sensing, 2019, 11(7): 765. doi: 10.3390/rs11070765
[7]	WEI Shunjun, ZENG Xiangfeng, QU Qizhe, et al. HRSID: A high-resolution SAR images dataset for ship detection and instance segmentation[J]. IEEE Access, 2020, 8: 120234–120254. doi: 10.1109/ACCESS.2020.3005861
[8]	ZHANG Tianwen, ZHANG Xiaoling, KE Xiao, et al. LS-SSDD-v1.0: A deep learning dataset dedicated to small ship detection from large-scale Sentinel-1 SAR images[J]. Remote Sensing, 2020, 12(18): 2997. doi: 10.3390/RS12182997
[9]	LEI Songlin, LU Dongdong, QIU Xiaolan, et al. SRSDD-v1.0: A high-resolution SAR rotation ship detection dataset[J]. Remote Sensing, 2021, 13(24): 5104. doi: 10.3390/rs13245104
[10]	HUANG Lanqing, LIU Bin, LI Boying, et al. OpenSARShip: A dataset dedicated to Sentinel-1 ship interpretation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018, 11(1): 195–208. doi: 10.1109/JSTARS.2017.2755672
[11]	LI Boying, LIU Bin, HUANG Lanqing, et al. OpenSARShip 2.0: A large-volume dataset for deeper interpretation of ship targets in Sentinel-1 imagery[C]. 2017 SAR in Big Data Era: Models, Methods and Applications, Beijing, China, 2017: 1–5.
[12]	HOU Xiyue, AO Wei, SONG Qian, et al. FUSAR-Ship: Building a high-resolution SAR-AIS matchup dataset of Gaofen-3 for ship detection and recognition[J]. Science China Information Sciences, 2020, 63(4): 140303. doi: 10.1007/s11432-019-2772-5
[13]	ZHANG Tianwen, ZHANG Xiaoling, LI Jianwei, et al. SAR ship detection dataset (SSDD): Official release and comprehensive data analysis[J]. Remote Sensing, 2021, 13(18): 3690. doi: 10.3390/rs13183690
[14]	European Space Agency[EB/OL]. https://www.esa.int/, 2022.
[15]	ITTVIS. ENVI-Image processing and analysis software solution[EB/OL]. https://www.ittvis.com/envi/, 2022.
[16]	PIE-SAR[EB/OL]. https://www.piesat.cn/product/PIE-SAR/index.html, 2022.
[17]	EVERINGHAM M, VAN GOOL L, WILLIAMS C K I, et al. The Pascal visual object classes (VOC) challenge[J]. International Journal of Computer Vision, 2010, 88(2): 303–338. doi: 10.1007/s11263-009-0275-4
[18]	HE Yishan, GAO Feo, WANG Jun, et al. Learning polar encodings for arbitrary-oriented ship detection in SAR images[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14: 3846–3859. doi: 10.1109/JSTARS.2021.3068530
[19]	RoLabelImg: Label rotated rect on images for training[EB/OL].https://github.com/cgvict/roLabelImg, 2022.
[20]	LIN T Y, MAIRE M, BELONGIE S, et al. Microsoft COCO: Common objects in context[C]. The 13th European Conference on Computer Vision, Zurich, Switzerland, 2014: 740–755.
[21]	REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[C]. The 28th International Conference on Neural Information Processing Systems, Montreal, Canada, 2015: 91–99.
[22]	REDMON J and FARHADI A. YOLO9000: Better, faster, stronger[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 6517–6525.
[23]	REDMON J and FARHADI A. YOLOv3: An incremental improvement[EB/OL]. https://arxiv.org/abs/1804.02767, 2018.
[24]	BOCHKOVSKIY A, WANG C Y, and LIAO H Y M. YOLOv4: Optimal speed and accuracy of object detection[EB/OL]. https://arxiv.org/abs/2004.10934, 2020.
[25]	LIU Wei, ANGUELOV D, ERHAN D, et al. SSD: Single shot MultiBox detector[C]. The 14th European Conference on Computer Vision, Amsterdam, The Netherlands, 2016: 21–37.
[26]	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]. 2017 IEEE International Conference on Computer Vision, Venice, Italy, 2017: 2999–3007.
[27]	ZHOU Xingyi, WANG Dequan, and KRÄHENBÜHL P. Objects as points[EB/OL]. http://arxiv.org/abs/1904.07850, 2019.
[28]	TIAN Zhi, SHEN Chunhua, CHEN Hao, et al. FCOS: Fully convolutional one-stage object detection[C]. 2019 IEEE/CVF International Conference on Computer Vision, Seoul, South Korea, 2019: 9626–9635.
[29]	DING Jian, XUE Nan, LONG Yang, et al. Learning RoI transformer for oriented object detection in aerial images[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 2844–2853.
[30]	XU Yongchao, FU Mingtao, WANG Qimeng, et al. Gliding vertex on the horizontal bounding box for multi-oriented object detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(4): 1452–1459. doi: 10.1109/TPAMI.2020.2974745
[31]	XIE Xingxing, CHENG Gong, WANG Jiabao, et al. Oriented R-CNN for object detection[C]. 2021 IEEE/CVF International Conference on Computer Vision, Montreal, Canada, 2021: 3500–3509.
[32]	HAN Jiaming, DING Jian, XUE Nan, et al. ReDet: A rotation-equivariant detector for aerial object detection[C]. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Montreal, Canada, 2021: 2785–2794.
[33]	HAN Jiaming, DING Jian, LI Jie, et al. Align deep features for oriented object detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5602511. doi: 10.1109/TGRS.2021.3062048
[34]	YANG Xue, YAN Junchi, FENG Ziming, et al. R3Det: Refined single-stage detector with feature refinement for rotating object[C]. Thirty-Fifth AAAI Conference on Artificial Intelligence, Vancouver, Canada, 2021.
[35]	GUO Zonghao, LIU Chang, ZHANG Xiaosong, et al. Beyond bounding-box: Convex-hull feature adaptation for oriented and densely packed object detection[C]. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 8788–8797.
[36]	AN Quanzhi, PAN Zongxu, LIU Lei, et al. DRBox-v2: An improved detector with rotatable boxes for target detection in SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(11): 8333–8349. doi: 10.1109/TGRS.2019.2920534
[37]	HU Shimin, LIANG Dun, YANG Guoye, et al. Jittor: A novel deep learning framework with meta-operators and unified graph execution[J]. Science China Information Sciences, 2020, 63(12): 222103. doi: 10.1007/s11432-020-3097-4
[38]	ZHOU Yue, YANG Xue, ZHANG Gefan, et al. MMRotate: A rotated object detection benchmark using PyTorch[EB/OL]. https://github.com/open-mmlab/mmrotate, 2022.
[39]	GOYAL P, DOLLÁR P, GIRSHICK R, et al. Accurate, large minibatch SGD: Training ImageNet in 1 hour[EB/OL]. https://arxiv.org/abs/1706.02677, 2018.
[40]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 770–778.