Mars-Bench S5Mars: Curiosity Mastcam Terrain Segmentation (9 classes, 1200 x 1200)

Provider
Mirali Purohit, Hannah Kerner et al. (Mars-Bench, Arizona State University)
License
CC-BY-4.0 (source)
Size
16 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/Mirali33/mb-s5mars)
Access
Free and public. No account with the data provider is needed.
DOI
10.48550/arXiv.2510.24010
Category
Astronomy & Space
Source
https://huggingface.co/datasets/Mirali33/mb-s5mars

About this dataset

Semantic segmentation of Curiosity rover Mastcam scenes at 1200 x 1200 pixels, from S5Mars (Zhang et al., 2022). Labels cover background, bedrock, hole, ridge, rock, rover, sand/soil, sky and track, a finer breakdown than MarsSeg that separates ridges, holes and rover hardware. Splits: 4,997 training, 200 validation, 800 test images. Used to train rover terrain classifiers for navigation and geologic context, and in Mars-Bench to test whether Earth-trained segmentation backbones transfer to Martian surface imagery. Parquet shards under data/ hold image, mask, width, height and class labels, with mapping.json for names; load with datasets.load_dataset("Mirali33/mb-s5mars"). About 2.8 GB. Part of Mars-Bench (Purohit et al., NeurIPS 2025 Datasets and Benchmarks track, arXiv:2510.24010), 20 standardized Mars datasets released by the authors under CC BY 4.0; cite Mars-Bench and Zhang et al. (2022), S5Mars.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("Mirali33/mb-s5mars")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: data/partition_train_0.01x_partition-00000-of-00001.parquet

Analyze this dataset without downloading it

On Scigantic, Mars-Bench S5Mars: Curiosity Mastcam Terrain Segmentation (9 classes, 1200 x 1200) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Astronomy & Space datasets