janruettinger / unsup3d_extended

This repository contains a Pytorch3D port of the code of the paper "Unsupervised Learning of Probably Symmetric Deformable 3D Objects from Images in the Wild" by Wu et al. You can find the original code here.

Differences compared to original version

• Pytorch3D instead of Neural Renderer is used as a differential renderer.
• The size of the albedo map (new: 128x128) and depth map (new: 32x32) have changed. Gradient update is unstable with original size of 64x64 for both maps.

Setup (with Anaconda)

conda env create -f environment.yml

Tested with Python 3.8.3.

Bug in Pytorch3d

You need to change the following lines in phong_shading() function in the file shading.py in the pytorch3d package. #colors = (ambient + diffuse) * texels + specular #colors = (ambient.unsqueeze(1).unsqueeze(1).unsqueeze(1) + diffuse) * texels + specula

Datasets

1. CelebA face dataset. Please download the original images (img_celeba.7z) from their website and run celeba_crop.py in data/ to crop the images.
2. Cat face dataset composed of Cat Head Dataset and Oxford-IIIT Pet Dataset (license). This can be downloaded using the script download_cat.sh provided in data/.
3. Dogs dataset (see https://github.com/JanRuettinger/dog_heads_dataset)

More coming soon