EgoForce: Forearm-Guided Camera-Space 3D Hand Pose from a Monocular Egocentric Camera

Project Page | Paper | HF Paper | Code | Demo

EgoForce is a monocular 3D hand reconstruction framework that recovers robust, absolute 3D hand pose and its position from the user's (camera-space) viewpoint. It operates across fisheye, perspective, and distorted wide-FOV camera models using a single unified network.

Repository contents

This repository contains EgoForce model assets and supporting files for egocentric hand and arm perception.

Path	Role	Provenance	License / notice
_DATA/detector.torchscript	2D hand bounding-box detector	WiLoR detector	Third-party asset; see WiLoR notice below
_DATA/mano/*	MANO hand model files	Official MANO distribution	Third-party asset; see MANO notice below
_DATA/epoch_460.pth	EgoForce hand-arm detector	EgoForce	First-party project asset
_DATA/model_weights.pth	EgoForce model	EgoForce	First-party project asset

Important licensing note

No single license applies to every file in this repository.

• The EgoForce code and model weights are released under the CC-BY-NC 4.0 license.
• _DATA/detector.torchscript is subject to the original WiLoR model license (CC BY-NC-ND 4.0).
• _DATA/mano/* remains subject to the official MANO non-commercial research license.

For detailed information, please refer to LICENSE.md.

Required citations

If you use this repository, please cite the corresponding EgoForce paper:

@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 ACM SIGGRAPH (Conference Track)},
      year={2026}
}

If you use the WiLoR detector or MANO assets bundled here, please cite their respective works:

@InProceedings{Potamias_2025_CVPR,
  author    = {Potamias, Rolandos Alexandros and Zhang, Jinglei and Deng, Jiankang and Zafeiriou, Stefanos},
  title     = {WiLoR: End-to-end 3D Hand Localization and Reconstruction in-the-wild},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2025}
}

@article{MANO:SIGGRAPHASIA:2017,
  title   = {Embodied Hands: Modeling and Capturing Hands and Bodies Together},
  author  = {Romero, Javier and Tzionas, Dimitrios and Black, Michael J.},
  journal = {ACM Transactions on Graphics, (Proc. SIGGRAPH Asia)},
  volume  = {36},
  number  = {6},
  year    = {2017}
}