Pixel2Mesh

Reimplemetation of Pixel2Mesh (ECCV 2018) by Team 19 of CS492(A), KAIST 2022 Spring Semester

Prerequisites

NVIDIA Driver

Check if already installed:

nvidia-smi

If not installed, follow the installation guide

Conda

Check if already installed:

conda --version

If not installed, follow miniconda installation guide

Setting Up the Environment

git clone https://github.com/sundoufu/Pixel2Mesh.git
cd Pixel2Mesh/
conda env create -f environment.yml
conda activate pytorch3d
pip install git+https://github.com/sundoufu/PyTorchEMD.git

Pretrained Checkpoints

1. Download from https://drive.google.com/drive/folders/1fXnnjMysnHf_vP3X6t2f3xv_9StFER_4?usp=sharing
2. Save them under Pixel2Mesh/checkpoints/

Dataset

1. Download ShapeNetP2M/ from https://drive.google.com/drive/folders/131dH36qXCabym1JjSmEpSQZg4dmZVQid
2. Save them under data/ (data/ShapeNetP2M)

Running the evaluation

# Inside Pixel2Mesh/
chmod +x ./eval.sh
# Run eval.sh in the background and redirect output logs to some files
nohup ./eval.sh 1>eval.out 2>eval.err &

# To see the progress real-time,
tail -f eval.err

Our Experiment Setting

These aren't necessaraily required, but for your information.

• Graphics: GEFORCE RTX 3090
• OS: Ubuntu 20.04.2 LTS (GNU/Linux 5.4.0-77-generic x86_64)