摘要

我们提供了一个新的实例级飞机数据库，包含1个实例飞机物体，型号为中国航空747。从搜索引擎上使用网络爬虫获得图片数据，同时在不同的物体遮挡和光照强度下对中国航空747模型进行了多视角的拍摄。该数据集包含共1661张1280*720的图像，和人工标注矩形框的xml文件（标注工具为LabelImg）。

Abstract

We provide a new instance-level plane dataset with one plane instance-level object whose type is Airchina 747. Obtain image data from search engines using web crawlers and take multi-view shots of Airchina 747 model under different object occlusions and light intensities. The dataset contains 1661 images with 1280*720 resolution and the xml files which are manually annotated with bounding boxes (The tool is LabelImg).

内容

· 包含1104张图像的训练集，558张图像的测试集

· 每张图像中包含单个或多个实例飞机物体

· 对实例飞机物体进行了矩形框标注的xml文件

· 拍摄设备：Canon EOS 80D, 18-135mmIS USM镜头

有关更多详细信息，请参阅数据链接中包含的README文件。

Contents

· Train dataset contains 1104 images and test dataset contains 558 images.

· Each image contains single or multiple plane instances.

· The xml files which plane instances are annotated with bounding boxes

· Equipment: Canon EOS 80D, 18-135mmIS USM.

For more details please see the README files that are included in the data links.

数据下载(Data)

BHID-P： bhid_p.zip

脚本

使用爬虫获得图片（python）:  get_image.zip

对图片进行归一化并重新命名（python）: normalization_and_rename.zip

Scripts

Get images by using crawler (python): get_image.zip

Normalize and rename images(python): normalization_and_rename.zip