Mars-Bench
收藏资源简介:
Mars-Bench是一个全面的基准测试,旨在使用轨道和地表图像系统地评估机器学习模型在广泛的火星相关任务上的性能。该基准包括20个数据集,涵盖分类、分割和目标检测任务,重点关注火星科学中常见的地质特征,如陨石坑、火山锥、巨石和霜。数据集已标准化,并提供预训练模型的基线评估,旨在促进火星科学领域机器学习模型的发展和比较。
Mars-Bench is a comprehensive benchmark designed to systematically evaluate the performance of machine learning models across a wide range of Mars-related tasks using orbital and surface images. This benchmark includes 20 datasets covering classification, segmentation and object detection tasks, focusing on geologic features common in Martian science, such as craters, volcanic cones, boulders and frost. The datasets have been standardized, and baseline evaluations with pre-trained models are provided, aiming to facilitate the development and comparison of machine learning models in the field of Martian science.

- 1Mars-Bench: A Benchmark for Evaluating Foundation Models for Mars Science Tasks亚利桑那州立大学计算与增强智能学院、地球与空间探索学院、喷气推进实验室 · 2025年



