Datasetpublic

nasa-ibm-ai4science/Sombench-WAC-Crater-Detection

SomBench Benchmark: Robbins Crater Detection, WAC Science theme: Impact processes Task: Object detection Dataset Summary An impact-crater object-detection benchmark built from the Robbins (2019) global lunar crater catalog, a manually compiled, near-complete census of lunar impact craters (≥ ~1–2 km). Catalog crater centers and diameters are converted to bounding boxes and packaged over LROC WAC visible tiles drawn from the pre-training corpus test split, in COCO… See the full description on the dataset page: https://huggingface.co/datasets/nasa-ibm-ai4science/Sombench-WAC-Crater-Detection.

sourceHugging Facecc-by-4.0updated 19d agoView on Hugging Face

0likes517downloads

Dataset Card

SomBench Benchmark: Robbins Crater Detection, WAC

Science theme: Impact processes

Task: Object detection

Dataset Summary

An impact-crater object-detection benchmark built from the [Robbins (2019)](https://doi.org/10.1029/2018JE005592) global lunar crater catalog, a manually compiled, near-complete census of lunar impact craters (≥ ~1–2 km). Catalog crater centers and diameters are converted to bounding boxes and packaged over LROC WAC visible tiles drawn from the pre-training corpus test split, in COCO format. The release contains 1,000 WAC VIS tiles (512 × 512 px, 100 m/pixel) split into train / validation / test.

Single object class: `crater`.

Dataset Structure

Directory layout

wac_craters_dataset/
├── README.md         # data source overview
├── images_tiff/      # 1,000 WAC VIS tiles (Tiff), 512 × 512 px, 100 m/px
├── metadata.parquet  # per-tile metadata + split assignments
├── train.json        # COCO annotations, train split
├── val.json          # COCO annotations, validation split
└── test.json         # COCO annotations, test split

Data files and splits

split	images	annotations	craters/img	median √area	incidence°	emission°
train	800	91,516	114.4	16.5	72.2	1.81
val	100	11,429	114.3	16.5	71.1	1.55
test	100	11,384	113.8	17.0	72.7	1.39

Images

—`images_tiff/`: Tiff version of the same tiles for out of the box data loading in terratorch.
—`metadata.parquet`: the parquet catalog of the tiles, their corresponding metadata (e.g. INCIDENCE_ANGLE), and their assigned dataset split (DATASET).
—`train.json / test.json / val.json`: the COCO annotation files for each split.

COCO schema

—Tiles are single-band .tif WAC VIS tiles (file_name), 512 × 512 px at 100 m/px.
—One category: {id: 1, name: "crater"}.
—Each annotation has polygon segmentation, bbox [x_min, y_min, width, height] (px), and area (px²).
—bbox is [x_min, y_min, width, height] in pixels.
—segmentation is a polygon approximating the (circular) crater outline.

Known Limitations

—Labels inherit the Robbins catalog's completeness limits (near-complete for diameters ≥ ~1-2 km); very small craters may be under-represented.
—The WAC visible tiles contain incidence angles between 60-80° apparent crater expression varies with lighting.

Citation

bibtex

@article{fraccaro2026lfm,
  title  = {Multimodal-Multiresolution Foundation Model for Lunar Remote Sensing},
  author = {Fraccaro, Paolo and Nyirjesy, Gabby and Szwarcman, Daniela and Patil, Himanshu
            and Gaur, Vishal and Lal, Rohit and Slank, Rachel A. and Dawson, Geoffrey
            and Debary, Hiyam and Dionelis, Nikolaos and Barker, Michael K. and Annex, Andrew
            and Viswanathan, Vishnu and Morse, Zachary and Schaefer, Ethan I. and Kumar, Ankur
            and Watson, Campbell D. and Dawson-Rigas, Rebekah I. and Maskey, Manil
            and Roy, Sujit and Ramachandran, Rahul and Bernab\'e-Moreno, Juan},
  year   = {2026}
  howpublished = {\url{https://huggingface.co/collections/nasa-ibm-ai4science/nasa-ibm-lunar-fm-and-downstream-models}}
}

@misc{sombench2026collection,
  author = {Patil, Himanshu and Nyirjesy, Gabby and Slank, Rachel A. and Gaur, Vishal
          and Szwarcman, Daniela and Fraccaro, Paolo and Dionelis, Nikolaos and Barker, Michael K.
          and Annex, Andrew and Viswanathan, Vishnu and Morse, Zachary and Schaefer, Ethan I.
          and Debary, Hiyam and Kumar, Ankur and Lal, Rohit and Dawson, Geoffrey
          and Watson, Campbell and Dawson-Rigas, Rebekah I. and Maskey, Manil
          and Bernab\'e-Moreno, Juan and Ramachandran, Rahul and Roy, Sujit},
  title        = {{SomBench}: Benchmark Dataset for Advancing Machine Learning in Lunar Science},
  year         = {2026},
  howpublished = {\url{https://huggingface.co/collections/nasa-ibm-ai4science/lunar-fm-ml-ready-benchmark-dataset-sombench}}
}

Data sources

The crater labels are derived from the Robbins (2019) global lunar crater catalog; please also cite the source catalog:

bibtex

@article{robbins2019craters,
  author  = {Robbins, Stuart J.},
  title   = {{A New Global Database of Lunar Impact Craters >1--2 km:
             1. Crater Locations and Sizes, Comparisons With Published
             Databases, and Global Analysis}},
  journal = {Journal of Geophysical Research: Planets},
  year    = {2019},
  volume  = {124},
  number  = {4},
  pages   = {871--892},
  doi     = {10.1029/2018JE005592}
}

Robbins, S. J. (2019). A new global database of lunar impact craters >1–2 km: 1. Crater locations and sizes, comparisons with published databases, and global analysis. Journal of Geophysical Research: Planets, 124(4), 871–892. <https://doi.org/10.1029/2018JE005592>

License

Released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.