metadata
license: apache-2.0
task_categories:
- image-to-3d
- depth-estimation
language:
- en
tags:
- computer-vision
- 3d-reconstruction
- multi-view-stereo
- depth-estimation
- camera-pose
- covisibility
- mapanything
pretty_name: MapAnything Dataset
size_categories:
- 100B<n<1T
MapAnything Dataset
Dataset Description
This dataset contains pre-computed metadata and covisibility matrices for supporting the MapAnything codebase. This metadata enables easy reproducible training and benchmarking for feed-forward 3D reconstruction tasks.
Please see our Data Processing README for more details.
Citation
If you use this dataset in your research, please cite our paper:
@inproceedings{keetha2025mapanything,
title={{MapAnything}: Universal Feed-Forward Metric {3D} Reconstruction},
author={Nikhil Keetha and Norman Müller and Johannes Schönberger and Lorenzo Porzi and Yuchen Zhang and Tobias Fischer and Arno Knapitsch and Duncan Zauss and Ethan Weber and Nelson Antunes and Jonathon Luiten and Manuel Lopez-Antequera and Samuel Rota Bulò and Christian Richardt and Deva Ramanan and Sebastian Scherer and Peter Kontschieder},
booktitle={arXiv},
year={2025}
}
License
This dataset is released under the Apache 2.0 License.
Links
- 🏠 Project Page: https://map-anything.github.io
- 💻 Code: GitHub Repository