DOORS（Dataset fOr bOuldeRs Segmentation）

定位在球形网格表面上的单个巨石的分段掩模。
检测小物体表面巨石的能力有利于基于视觉的应用，如关键操作和导航期间的危险检测。由于各种各样的不规则形状，巨石种群的特征以及照明条件的快速变化，该任务具有挑战性。此外，这些应用程序缺乏公开可用的标记数据集，这阻碍了有关数据驱动算法的研究。为了应对这些挑战，已经设计了巨石分割 (门) 数据集。数据集被认为对于 (但不限于) 巨石识别、质心回归、分割和导航应用是有用的。数据集分为两组:
回归: 包含定位在球形网格表面上的单个巨石的4个分割的图像，蒙版和标签。它可用于执行导航、巨石识别、分割和质心回归。
分割: 包含2个数据集的图像、蒙版和标签: DS1和ds2。DS1由回归数据集的相同图像制成，但专门为分割而设计。DS2由多个巨石实例出现在Didymos小行星模型表面上的图像组成

A segmented mask of a single boulder located on the surface of a spherical grid.
The ability to detect boulders on small object surfaces is beneficial for vision-based applications such as hazard detection during critical operations and navigation. Owing to the wide range of irregular shapes, diverse characteristics of boulder populations, and rapid fluctuations in illumination conditions, this task poses significant challenges. Furthermore, the lack of publicly available labeled datasets for these applications has hindered research into data-driven algorithms.
To address these challenges, the Boulder Segmentation (Door) Dataset has been developed. This dataset is useful for, but not limited to, boulder recognition, centroid regression, segmentation, and navigation applications.
The dataset is divided into two groups:
1. Regression: Contains 4 segmented images, masks, and labels of a single boulder positioned on the spherical grid surface. It can be used for navigation, boulder recognition, segmentation, and centroid regression tasks.
2. Segmentation: Contains images, masks, and labels for two datasets: DS1 and DS2. DS1 is constructed from the same images as the Regression dataset, and is specifically designed for segmentation tasks. DS2 consists of images showing multiple boulder instances on the surface of the Didymos asteroid model.