TianhangCheng7/DuplicateWeight

Weights for Structure from Duplicates (SfD)

Pretrained checkpoints used by the SfD preprocessing pipeline (Tianhang-Cheng/SfD). They are hosted here instead of in the git repository to keep clones small.

file	size	what it is
keypoint_matching/superpoint_v1.pth	5 MB	SuperPoint detector/descriptor
keypoint_matching/superglue_indoor.pth	46 MB	SuperGlue matcher, indoor weights
keypoint_matching/superglue_outdoor.pth	46 MB	SuperGlue matcher, outdoor weights

Usage

From a clone of the repo, all three land in preprocess/keypoint_matching/weights/:

python download_assets.py --weights

The pipeline also fetches them lazily the first time the matcher is constructed, so this step is optional.

Directly:

from huggingface_hub import hf_hub_download

hf_hub_download("TianhangCheng7/DuplicateWeight", "keypoint_matching/superpoint_v1.pth")

License / attribution

SuperPoint and SuperGlue were released by Magic Leap, Inc. and are mirrored here under the original SuperGluePretrainedNetwork terms, which allow non-commercial research use only. Please cite the original works:

@inproceedings{detone2018superpoint,
  title={SuperPoint: Self-Supervised Interest Point Detection and Description},
  author={DeTone, Daniel and Malisiewicz, Tomasz and Rabinovich, Andrew},
  booktitle={CVPR Deep Learning for Visual SLAM Workshop},
  year={2018}
}
@inproceedings{sarlin2020superglue,
  title={SuperGlue: Learning Feature Matching with Graph Neural Networks},
  author={Sarlin, Paul-Edouard and DeTone, Daniel and Malisiewicz, Tomasz and Rabinovich, Andrew},
  booktitle={CVPR},
  year={2020}
}

If you use SfD:

@inproceedings{cheng2023structure,
  title={Structure from Duplicates: Neural Inverse Graphics from a Pile of Objects},
  author={Cheng, Tianhang and Ma, Wei-Chiu and Guan, Kaiyu and Torralba, Antonio and Wang, Shenlong},
  booktitle={Thirty-seventh Conference on Neural Information Processing Systems},
  year={2023}
}