kuan-wei-tseng/artnv

Artistic Style Novel View Synthesis Based on A Single Image (CVPRW2022)

ArtNV

This is the official implementation of "Artistic Style Novel View Synthesis Based on A Single Image" (CVPR 2022 Workshop).

Note: ⚠️ This project is still under construction. The code will be ready in mid June, 2022.

Installation

Our codes are tested on the following environment:

• Python=3.8
• PyTorch=1.8.0
• PyTorch3D = 0.6.2

1. Create a new virtual environment with conda.

conda create -n artnv python=3.8
conda activate artnv

2. Install PyTorch with CUDA support.

Remember to check the compatbility between Python, PyTorch, and Pytorch3D. Do not use PyTorch 1.8.3 (LTS) as it is not compatbible with PyTorch3D.

conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge

3. Install PyTorch3D

Follow the installation guideline in the official documents.

pip install scikit-image matplotlib imageio plotly opencv-python
pip install black usort flake8 flake8-bugbear flake8-comprehensions
conda install -c fvcore -c iopath -c conda-forge fvcore iopath
conda install -c bottler nvidiacub
conda install pytorch3d -c pytorch3d

4. Clone and setup the project. Clone the project with all third-party submodules. Run setup.sh to download models for view synthesis and style transfer.

git clone --recurse-submodules https://github.com/Kuan-Wei-Tseng/ArtNV.git
cd ArtNV
sh setup.sh

Reference

If you find this code helpful, please cite

@InProceedings{Tseng_2022_CVPRW,
    author    = {Tseng, Kuan-Wei and Lee, Yao-Chih and Chen, Chu-Song},
    title     = {Artistic Style Novel View Synthesis Based on A Single Image},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
    month     = {June},
    year      = {2022},
}