AstroVision

该数据集名为AstroVision基准，它通过标准指标如精确度、召回率和准确度，对每对图像上计算出的特征性能进行评估。此外，该数据集还采用曲线下面积（AUC）来评估从假定匹配中计算出的相对姿态的质量，这一评估是基于标准化累积姿态误差曲线进行的。该任务的目的是进行特征匹配评估。

This dataset, named the AstroVision benchmark, evaluates the performance of features computed on image pairs using standard metrics including precision, recall, and accuracy. Furthermore, it uses the Area Under the Curve (AUC) to assess the quality of relative poses calculated from putative matches, with the evaluation conducted based on the standardized cumulative pose error curve. The goal of this task is to evaluate feature matching performance.