# Mars-Bench Boulder Detection, MRO HiRISE (YOLO, COCO and Pascal VOC boxes)

## 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: 10 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/Mirali33/mb-boulder_det
- 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-boulder_det
- Page: https://scigantic.com/archive/e1d88f15-5299-5e7b-a561-c1af2e7d7797
- Keywords: Astronomy & Space, Parquet, Hugging Face

## About

Object detection of boulders on Mars in 500 x 500 pixel MRO HiRISE tiles, with bounding boxes derived from the boulder outlines of Prieur et al. (2023). One class (boulder); splits are 39 training, 6 validation and 4 test tiles plus 5 to 50 percent training fractions, a deliberately low-data detection task with many small, crowded objects per image. Annotations ship in YOLO format (data_yolo.yaml) with COCO and Pascal VOC versions included, so the set plugs into Ultralytics YOLO, Detectron2 or torchvision detectors. The same tiles as masks are in mb-boulder_seg. Parquet shards under data/; load with datasets.load_dataset("Mirali33/mb-boulder_det"). About 30 MB. 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 Prieur et al. (2023).

## Load the data

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

## Data structure

File types in sampled files: Parquet

Example file: `data/0.10x_partition-00000-of-00001.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/e1d88f15-5299-5e7b-a561-c1af2e7d7797
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/e1d88f15-5299-5e7b-a561-c1af2e7d7797
- Over MCP: `get_schema_card` with id `e1d88f15-5299-5e7b-a561-c1af2e7d7797` at https://api.scigantic.com/mcp

## Related topics

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