Self-Supervised 3D Mesh Reconstruction from Single Images
Lists
Only you see this
Abstract
Recent single-view 3D reconstruction methods recon- struct object’s shape and texture from a single image with only 2D image-level annotation. However, without ex- plicit 3D attribute-level supervision, it is still difficult to achieve satisfying reconstruction accuracy. In this paper, we propose a Self-supervised Mesh Reconstruction (SMR) approach to enhance 3D mesh attribute learning process. Our approach is motivated by observations that (1) 3D at- tributes from interpolation and prediction should be consis- tent, and (2) feature representation of landmarks from all images should be consistent. By only requiring silhouette mask annotation, our SMR can be trained in an end-to- end manner and generalizes to reconstruct natural objects of birds, cows, motorbikes, etc. Experiments demonstrate that our approach improves both 2D supervised and unsu- pervised 3D mesh reconstruction on multiple datasets. We also show that our model can be adapted to other image synthesis tasks, e.g., novel view generation, shape trans- fer, and texture transfer, with promising results.
BibTeX
@inproceedings{hu2021self,
author = {Tao Hu},
title = {{Self-Supervised 3D Mesh Reconstruction from Single Images}},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year = {2021},
url = {https://openreview.net/forum?id=JQjUHHp46ZH},
}Loading related content...