DOTA v2.0

该数据集名为DOTA v2.0，是遥感领域旋转目标检测的一个基准数据集。它包含了来自多种来源的图像，覆盖了众多类别。在训练集中，包含了1,830张图像和268,627个实例。验证集和测试集则分别具有特定的图像数量和实例数。总体而言，该数据集共包含了11,268张图像和1,793,658个实例，其任务是进行旋转目标检测。

This dataset, named DOTA v2.0, is a benchmark dataset for rotated object detection in the remote sensing domain. It comprises images from multiple sources and covers a wide range of object categories. The training split includes 1,830 images and 268,627 object instances, while the validation and test splits respectively have specified counts of images and instances. Overall, the full dataset contains a total of 11,268 images and 1,793,658 instances, with its core task being rotated object detection.

搜集汇总

背景与挑战

背景概述

DOTA v2.0是一个用于航空图像中目标检测的大规模基准数据集，提供了多种任务和评估服务器。使用该数据集时需注意格式要求，避免空文件导致的评估错误。