Mars-Bench: mb-boulder_seg

This is a binary segmentation dataset focused on segmenting boulders on the Martian surface using high-resolution orbital imagery from the HiRISE camera. The dataset comprises manually annotated binary masks indicating the presence or absence of boulders within each image. Boulders were annotated by planetary scientists using precise polygon outlines, ensuring high-quality labels. This is one of the smallest datasets in Mars-Bench with only tens of samples, and that makes it challenging for the computer vision community. To explore all Mars-Bench resources, visit our project website: https://mars-bench.github.io/本数据集为二值语义分割数据集，旨在借助高分辨率成像科学实验相机（HiRISE）获取的高分辨率轨道影像，对火星表面的巨石开展分割任务。该数据集包含人工标注的二值掩码，用于标注每张影像内巨石的存在与否。行星科学家采用精准的多边形轮廓对巨石进行标注，以此确保标注标签的高质量水准。本数据集是Mars-Bench中规模最小的数据集之一，仅包含数十个样本，这对计算机视觉学界而言颇具挑战性。 若需探索Mars-Bench的全部资源，请访问我们的项目官网： https://mars-bench.github.io/