We introduce a simple and effective network architecture for monocular 3D hand pose estimation consisting of an image encoder followed by a mesh convolutional decoder that is trained through a direct 3D hand mesh reconstruction loss. We train our network by gathering a large-scale dataset of hand action in YouTube videos and use it as a source of weak supervision. Our weakly-supervised mesh convolutions-based system largely outperforms state-of-the-art methods, even halving the errors on the in the wild benchmark. The dataset and additional resources are available at https://arielai.com/mesh_hands.
@article{arxiv.2004.01946,
title = {Weakly-Supervised Mesh-Convolutional Hand Reconstruction in the Wild},
author = {Dominik Kulon and Riza Alp Güler and Iasonas Kokkinos and Michael Bronstein and Stefanos Zafeiriou},
journal= {arXiv preprint arXiv:2004.01946},
year = {2020}
}
Comments
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2020). Additional resources: https://arielai.com/mesh_hands