EgoForce Evaluation Data

Project Page | Paper | Code | Demo

This repository contains the evaluation data used for EgoForce: Forearm-Guided Camera-Space 3D Hand Pose from a Monocular Egocentric Camera.

It includes official evaluation datasets together with EgoForce-specific processed artifacts:

• forearm parameters
• hand visibility and occlusion metadata
• filtered hand-present indices
• EgoForce HOT3D validation split

The datasets currently present in this repository are:

• ARCTIC
• H2O
• HOT3D

Usage

To download the datasets, you can use the provided script from the official repository:

bash scripts/download_datasets.sh --data-root /path/to/datasets

After download, update settings.py so config.DATASET.DIR points to your dataset root.

Notice

This repository does not grant a new standalone license over the included data.

• The underlying datasets remain subject to their official upstream licenses and terms of use.
• EgoForce processing does not remove, relax, or replace those upstream restrictions.

See LICENSE for the composite provenance and redistribution notice.

Included Datasets

Dataset	Official paper / project	Official license source	Official license summary	EgoForce-specific processing in this repository
ARCTIC	ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulation	ARCTIC license	Non-commercial scientific research, non-commercial education, and non-commercial artistic use only. Commercial use is prohibited without separate permission.	EgoForce uses ARCTIC as an evaluation dataset and adds processed forearm annotations, hand masks, hand-joint occlusion metadata, filtered hand-present indices, and evaluation indexing files.
H2O	H2O: Two Hands Manipulating Objects for First Person Interaction Recognition	H2O terms of use	Academic use only. No commercial use. No transfer to any third party.	EgoForce uses H2O as an evaluation dataset and adds processed hand masks, and forearm annotations, filtered hand-present indices.
HOT3D	HOT3D: Hand and Object Tracking in 3D from Egocentric Multi-View Videos	HOT3D license agreement	Different official licenses for different asset types (CC BY-SA, CC BY-NC-SA).	EgoForce uses HOT3D as an evaluation dataset and adds hand bounding boxes, occlusion metadata, filtered hand-present indices, and a custom validation split.

Citation Requirements

If you use this repository or the processed evaluation setup, cite the corresponding EgoForce paper and cite the original datasets you use.

EgoForce

@inproceedings{millerdurai2026egoforce,
      title={EgoForce: Forearm-Guided Camera-Space 3D Hand Pose from a Monocular Egocentric Camera},
      author={Millerdurai, Christen and Wang, Shaoxiang and Xie, Yaxu and Golyanik, Vladislav and Stricker, Didier and Pagani, Alain},
      booktitle={Proceedings of the SIGGRAPH 2026 Conference Papers},
      year={2026}
}

ARCTIC

@inproceedings{fan2023arctic,
  title = {{ARCTIC}: A Dataset for Dexterous Bimanual Hand-Object Manipulation},
  author = {Fan, Zicong and Taheri, Omid and Tzionas, Dimitrios and Kocabas, Muhammed and Kaufmann, Manuel and Black, Michael J. and Hilliges, Otmar},
  booktitle = {Proceedings IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2023}
}

H2O

@InProceedings{Kwon_2021_ICCV,
  author = {Kwon, Taein and Tekin, Bugra and St\"uhmer, Jan and Bogo, Federica and Pollefeys, Marc},
  title = {H2O: Two Hands Manipulating Objects for First Person Interaction Recognition},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  month = {October},
  year = {2021},
  pages = {10138-10148}
}

HOT3D

@article{banerjee2024hot3d,
  title={{HOT3D}: Hand and Object Tracking in {3D} from Egocentric Multi-View Videos},
  author={Banerjee, Prithviraj and Shkodrani, Sindi and Moulon, Pierre and Hampali, Shreyas and Han, Shangchen and Zhang, Fan and Zhang, Linguang and Fountain, Jade and Miller, Edward and Basol, Selen and Newcombe, Richard and Wang, Robert and Engel, Jakob Julian and Hodan, Tomas},
  journal={CVPR},
  year={2025}
}