European Conference on Computer Vision (ECCV 2022). [Arxiv, PDF, Supp, Project]
We propose a novel image-conditioned neural implicit field, which can leverage 2D supervisions from GAN-generated multi-view images and perform the single-view reconstruction of generic objects. Firstly, a novel offline StyleGAN-based generator is presented to generate plausible pseudo images with full control over the viewpoint. Then, we propose to utilize a neural implicit function, along with a differentiable renderer to learn 3D geometry from pseudo images with object masks and rough pose initializations. To further detect the unreliable supervisions, we introduce a novel uncertainty module to predict uncertainty maps, which remedy the negative effect of uncertain regions in pseudo images, leading to a better reconstruction performance. The effectiveness of our approach is demonstrated through superior single-view 3D reconstruction results of generic objects.
The code is developed with
- Python 3.7
- Pytorch 18
- Cuda 11.1
-
Please follow here to prepare the training data.
-
Train the model:
cd code python training/exp_runner.py --conf ./confs/car.conf
@inproceedings{liu2022gansvr,
title={2D GANs Meet Unsupervised Single-View 3D Reconstruction},
author={Liu, Feng and Liu, Xiaoming},
booktitle={ECCV},
year={2022}}
Our implementation is heavily built upon idr. If you find our work useful, please also consider to cite this paper.
For questions feel free to post here or drop an email to - liufeng6@msu.edu