DOTA 2.0

该数据集是针对航拍图像中定向目标检测的最大数据集之一，包含了2,806张图像和188,282个实例，覆盖了15个常见类别。所有图像都被划分为600×600像素的子图像，并且每个子图像之间有150像素的重叠。该数据集规模庞大，所面临的任务是定向目标检测。

This dataset is among the largest available datasets for oriented object detection in aerial imagery. It contains 2,806 images and 188,282 instances across 15 common categories. All original images are split into 600×600 pixel sub-images, with a 150-pixel overlap between adjacent sub-images. As a large-scale dataset, it is tailored for the task of oriented object detection.

搜集汇总

背景与挑战

背景概述

DOTA 2.0是一个航空图像目标检测的大规模数据集，包含来自Google Earth、GF-2和JL-1卫星的RGB和灰度图像，标注采用定向边界框（OBB）格式，涵盖18个物体类别。数据集仅限学术使用，禁止商业用途。