# Mars-Bench Crater Morphology Segmentation, Mars Odyssey THEMIS (Layered, Buried, Secondary)

## Facts

- Provider: Mirali Purohit, Hannah Kerner et al. (Mars-Bench, Arizona State University)
- License: CC-BY-4.0 (https://arxiv.org/html/2510.24010v1)
- Size: 15 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/Mirali33/mb-crater_multi_seg
- Access: public, readable without an account
- DOI: https://doi.org/10.48550/arXiv.2510.24010
- Category: Astronomy & Space
- Source: https://huggingface.co/datasets/Mirali33/mb-crater_multi_seg
- Page: https://scigantic.com/archive/07ba1e6c-2f8b-5de7-a385-d1cef4437c80
- Keywords: Astronomy & Space, image segmentation, Parquet, Hugging Face

## About

Multi-class segmentation of Martian impact craters by morphology in 512 x 512 pixel Mars Odyssey THEMIS infrared tiles, with masks built by the Mars-Bench authors from the Lagain et al. (2021) crater database. Five classes: background, layered ejecta crater, buried crater, secondary crater, and other crater. Splits are 3,600 training, 900 validation and 900 test tiles, with 1 to 50 percent training fractions; metadata.zip holds tile positions. Distinguishing layered-ejecta craters (linked to subsurface ice) and secondaries (which bias crater-count ages) is the science use. Same tiles as mb-crater_binary_seg. Parquet shards under data/ (image, mask, width, height, class labels) with mapping.json; load with the datasets library. About 1.9 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 Lagain et al. (2021).

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("Mirali33/mb-crater_multi_seg")
print(ds)
```

## Data structure

File types in sampled files: Parquet

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

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/07ba1e6c-2f8b-5de7-a385-d1cef4437c80
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/07ba1e6c-2f8b-5de7-a385-d1cef4437c80
- Over MCP: `get_schema_card` with id `07ba1e6c-2f8b-5de7-a385-d1cef4437c80` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Image segmentation datasets](https://scigantic.com/topics/image-segmentation)
- [Astronomy & Space datasets](https://scigantic.com/topics/astronomy-and-space)
