航空遥感数据集

Posted 2020-12-19 2008nmj

原文：https://arxiv.org/pdf/1711.10398.pdf

官网：https://captain-whu.github.io/DOTA/

dataset：https://captain-whu.github.io/DOTA/dataset.html

Image Source and Usage License

The images of in DOTA-v1.0 dataset are manily collected from the Google Earth, some are taken by satellite JL-1, the others are taken by satellite GF-2 of the China Centre for Resources Satellite Data and Application.

Use of the images from Google Earth must respect the corresponding terms of use: "Google Earth" terms of use.

All images and their associated annotations in DOTA can be used for academic purposes only, but any commercial use is prohibited.

Object Category

The object categories in DOTA-v1.0 include: plane, ship, storage tank, baseball diamond, tennis court, basketball court, ground track field, harbor, bridge, large vehicle, small vehicle, helicopter, roundabout, soccer ball field and swimming pool.

Annotation format

In the dataset, each instance‘s location is annotated by a quadrilateral bounding boxes, which can be denoted as "x1, y1, x2, y2, x3, y3, x4, y4" where (xi, yi) denotes the positions of the oriented bounding boxes‘ vertices in the image. The vertices are arranged in a clockwise order. The following is the Visualization of adopted annotation method. The yellow point represents the starting point. which refers to: (a) top left corner of a plane, (b) top left corner of a large vehicle diamond, (c) the center of sector-shaped baseball.

Except the annotation of location, category label is assigned for each instance, which comes from one of the above 15 selected categories, and meanwhile a difficult label is provided which indicates whether the instance is difficult to be detected(1 for difficult, 0 for not difficult). Annotations for an image are saved in a text file with the same file name. At the first line, ‘imagesource‘(from GoogleEarth, GF-2 or JL-1) is given. At the second line, ’gsd’(ground sample distance, the physical size of one image pixel, in meters) is given. Note if the ‘gsd‘ is missing, it is annotated to be ‘null‘. From third line to last line in annotation text file, annotation for each instance is given. The annotation format is:

					
‘imagesource‘:imagesource
‘gsd‘:gsd
x1, y1, x2, y2, x3, y3, x4, y4, category, difficult
x1, y1, x2, y2, x3, y3, x4, y4, category, difficult
...						
					
				

(a)

(b)

(c)

 

Development kit

The Development kit provide the following function

• Load and visulize the data.
• Evaluate the result.
• Split and merge the data.

 

Trained Models

The following are the codes and models for DOTA.

• Object Detection Benchmarks for Aerial Images
• RoI Transformer
• Faster R-CNN OBB
• SSD and R-FCN
• YOLO-v2

 

Data Download

You can download DOTA-v1.0 from either Baidu Drive or Google Drive, according to your network connections.

• DOTA-v1.0 on Baidu Drive: Training set, Validation set, Testing images
• DOTA-v1.0 on Google Drive: Training set, Validation set, Testing images

 

 

 

News

• 2020-01-31 We release the code of new benchmarks. New
• 2019-12-31 iSAID Evaluation server is online. New
• 2019-06-14 "A large-scale dataset for instance segmentation in aerial images" (iSAID) has been released. New
• 2019-06-14 The code of “Learning RoI Transformer for Detecting Oriented Objects in Aerial Images” has released. New
• 2019-03-05 DOTA-v1.5 has released. New
• 2019-02-25 Our work “Learning RoI Transformer for Detecting Oriented Objects in Aerial Images” has been accepted by CVPR’2019. New
• 2018-08-19 We updated the leaderboard.
• 2018-06-27 You can make comments on the evaluation server page now.
• 2018-06-24 All the trained models mentioned in paper were released. You can find it in the Dataset page.
• 2018-04-30 The code of Faster-RCNN OBB is released.
• 2018-03-20 Fix little bug on gsd of annotation.
• 2018-03-16 The registration for ODAI is open now.
• 2018-03-14 We updated the results on the baseline algorithms in results page.
• 2018-03-8 DOTA development kit ia available now. It‘s helpful to play on DOTA!
• 2018-02-19 The article of DOTA has been accepted by CVPR‘2018.
• 2018-02-08 ODAI：a contest of object detection in aerial images on ICPR‘2018, is now open!
• 2018-01-27 A problem of annotations is fixed and a new-version annotation has been released.
• 2018-01-26 DOTA-v1.0 released with all images and oriented bounding box annotations for training and vallidation!

Description

Dota is a large-scale dataset for object detection in aerial images. It can be used to develop and evaluate object detectors in aerial images. We will continue to update DOTA, to grow in size and scope and to reflect evolving real-world conditions. For the DOTA-v1.0, as described in the paper, it contains 2806 aerial images from different sensors and platforms. Each image is of the size in the range from about 800 × 800 to 4000 × 4000 pixels and contains objects exhibiting a wide variety of scales, orientations, and shapes. These DOTA images are then annotated by experts in aerial image interpretation using 15 common object categories. The fully annotated DOTA images contains 188, 282 instances, each of which is labeled by an arbitrary (8 d.o.f.) quadrilateral.

For more details, refer to the arXiv preprint of DOTA.

Dota是一种用于航空图像目标检测的大规模数据集。它可以用来开发和评估航空图像中的目标探测器。我们将继续更新DOTA，以扩大规模和范围，并反映不断变化的现实世界条件。如本文所述，对于DOTA-v1.0，它包含来自不同传感器和平台的2806幅航空图像。每个图像的大小在800×800到4000×4000像素之间，并且包含显示各种比例、方向和形状的对象。这些DOTA图像然后由航空图像解释专家使用15种常见的物体类别进行注释。完整注释的DOTA图像包含188282个实例，每个实例都由任意（8d.o.f.）四边形标记。
有关详细信息，请参阅DOTA的arXiv预印本。

Examples of Annotated Images

Citation

If you make use of the DOTA dataset, please cite our following paper:

@InProceedings{Xia_2018_CVPR,
author = {Xia, Gui-Song and Bai, Xiang and Ding, Jian and Zhu, Zhen and Belongie, Serge and Luo, Jiebo and Datcu, Mihai and Pelillo, Marcello and Zhang, Liangpei},
title = {DOTA: A Large-Scale Dataset for Object Detection in Aerial Images},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2018}
}
	

@InProceedings{Ding_2019_CVPR,
author = {Jian Ding, Nan Xue, Yang Long, Gui-Song Xia, Qikai Lu},
title = {Learning RoI Transformer for Detecting Oriented Objects in Aerial Images},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}
	

Communicate

For any problem you have in using DOTA or ODAI, you can join the WeChat group and communicate.