Learning monocular 3D reconstruction of articulated categories from motion (ACFM)

Filippos Kokkinos and Iasonas Kokkinos

Paper Project Page

Requirements

• Python 3.6+
• PyTorch 1.7.0
• PyTorch3D 0.3.0, there are some breaking changes with more recent versions.
• cuda 11.0

For setup and installation refer to docs/install.md instructions.

Setup Evaluation and Training

We have separate folder for video and monocular based training.

• To train from scratch follow instructions for monocular training and video training.
• For evaluation please follow instructions for birds and quadrupeds.

Citation

If you find the code useful for your research, please consider citing:-

@InProceedings{Kokkinos_2021_CVPR,
               author = {Kokkinos, Filippos and Kokkinos, Iasonas},
               title = {Learning monocular 3D reconstruction of articulated categories from motion},
               booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
               month = {June},
               year = {2021}
}

Acknowledgements

This code repository uses code from CMR and CSM repos.

Contact

For questions feel free to contact me at filippos.kokkinos[at]ucl.ac.uk .