A 3D mesh can be represented by Images, Points, Voxels or Mesh. In this code we are trying to compare between these representations and analyze the tradeoffs between them. We can compare them on various tasks.
Tasks:
- Capturing details:
Question: Which representation would be best for capturing details?
Dataset Used: Co-Locating Style-Defining Elements on 3D Shapes
This implementation uses Pytorch.
Code has been tested on Pytorch 0.3.1 and Python 2.7.14
## Download the repository
git clone https://github.com/nitinagarwal/style_clustering.git
cd stlye_clustering
## Create a virtual enviroment
virtualenv .style_clustering_env
source .style_clutering_env/bin/activate
## Install the required packages
pip install -r requirements.txt
# You are Done!
Please read the train.py to start training
Our code is released under MIT License (see License file for details).
Please contact Nitin Agarwal if you have any questions or comments.