Pix2Vox with Silhouette Guidance(contributor:Zijie Tan,Zepeng Xiao,Shiyu Xiu)

Datasets

We use the ShapeNet datasets in our experiments, which are available below:

• ShapeNet rendering images: http://cvgl.stanford.edu/data2/ShapeNetRendering.tgz
• ShapeNet voxelized models: http://cvgl.stanford.edu/data2/ShapeNetVox32.tgz

Steps to run our code

Clone the Code Repository and switch to vox_renderer branch

git clone https://github.com/ZjTan4/Pix2Vox_414.git cd Pix2Vox_414

if you want to run our silhouette guided implementation, stay at main branch or go to vox_renderer branch,
they share exactly the same implementation, (optional)run:

git switch vox_renderer

if you want to run original pix2vox implementation in order to compare with own approach, run:

git switch baseline

Install Python Denpendencies

pip install -r requirements.txt

In addition, we also used torch, torchvision with cuda and pytorch3d. To install the correct version based on your hardware, please follow the instruction:

Instruction about installing torchvision iwth cuda:
https://pytorch.org/get-started/locally/

Instruction about installing pytorch3d:
- https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md

Update Settings in config.py

You need to update the file path of the datasets downloaded from previous section:

__C.DATASETS.SHAPENET.TAXONOMY_FILE_PATH    = '/path/ShapeNet.json'
__C.DATASETS.SHAPENET.RENDERING_PATH        = '/path/to/Datasets/ShapeNet/ShapeNetRendering/%s/%s/rendering/%02d.png'
__C.DATASETS.SHAPENET.VOXEL_PATH            = '/path/to/Datasets/ShapeNet/ShapeNetVox32/%s/%s/model.binvox'

How To Run the code

To train the model, run the command:

python3 runner.py

To test the model:

python3 runner.py --test --weights=/path/to/pretrained/model.pth

License

This project is open sourced under MIT license.