SAR Aircraft Detection Dataset

Data Description

Since there is no publicly available aircraft detection dataset in SAR images, we collect and construct an aircraft slice dataset to investigate the detection performance of our method. We name the SAR aircraft detection dataset SADD and make it public to facilitate the research of other scholars. The SADD is collected from the German TerraSAR-X satellite, which working in x-band and HH polarization mode with image resolutions ranging from 0.5m to 3m. The ground truth of aircraft are manually annotated by SAR ATR experts according to the priori knowledge and corresponding optical images. After cropping the large images, 2966 nonoverlapped 224×224 slices are collected with 7835 aircraft targets, of which structures, outlines and main components are clear. The distributions of bounding boxes’ sizes in pixels and aspect ratios are shown in (a), (b), respectively. Aircraft targets in SADD have various sizes, and there are a large number of small-size targets.

Example

The target background of SADD is relatively complex, including various scenes such as airport runway, airport apron, and civil aviation airport. The negative samples are mostly around the airport, including open space and forest, etc. The picture shows the sample images in SADD.

Label format

We use yolo-format for our labels, as shown bellow:

<object-class> <x_center> <y_center> <width> <height>

Where:

• <object-class> - integer object number from 0 to (classes-1)

• <x_center> <y_center> <width> <height> - float values relative to width and height of image, it can be equal from (0.0 to 1.0]

• for example: <x> = <absolute_x> / <image_width> or <height> = <absolute_height> / <image_height>

• attention: <x_center> <y_center> - are center of rectangle (are not top-left corner)

  For example for img1.jpg you will be created img1.txt containing:

  0 0.716797 0.395833 0.216406 0.147222
  0 0.687109 0.379167 0.255469 0.158333
  0 0.420312 0.395833 0.140625 0.166667
  

Data download

BaiduYun:

    https://pan.baidu.com/s/11SBfMkGszu3Lr_Pe1lIEkg
    Extraction Code: d2uw

If you use this data for your research, please consider citing:

@ARTICLE{9761751,  
author={Zhang, Peng and Xu, Hao and Tian, Tian and Gao, Peng and Li, Linfeng and Zhao, Tianming and Zhang, Nan and Tian, Jinwen},  
journal={IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},   
title={SEFEPNet: Scale Expansion and Feature Enhancement Pyramid Network for SAR Aircraft Detection With Small Sample Dataset},   
year={2022},  
volume={15},  
number={},  
pages={3365-3375},  
doi={10.1109/JSTARS.2022.3169339}}