By: Chiyu "Max" Jiang, Dana Lansigan, Philip Marcus, Matthias Niessner
[Project Website] [Paper]
This repository is based on paper: DDSL: Deep Differential Simplex Layer for Neural Networks. The project webpage presents an overview of the project.
In this project, we present a novel neural network layer that performs differentiable rasterization of arbitrary simplex-mesh-based geometrical signals (e.g., point clouds, line mesh, triangular mesh, tetrahedron mesh, polygon and polyhedron) of arbitrary dimensions. We further provide examples of incorporating the DDSL into neural networks for tasks such as polygonal image segmentation and neural shape optimization (for MNIST digits and airfoils).
Our deep learning code base is written using PyTorch 1.0 in Python 3, in conjunction with standard ML packages such as Numpy. PyTorch version > 1.0 is required.
We provide an efficient natively PyTorch-based implementation of the DDSL. Detailed documentation for APIs can be found in ddsl/ddsl.py. For examples on using the DDSL implementation for rasterizing a given input mesh, refer to the jupyter notebooks in the folder examples.
To replicate the experiments in our paper, please refer to codes in the experiments folder. Detailed instructions for each experiment can be found in
If you find our code useful for your work, please consider citing our paper:
@misc{jiang2019ddsl,
title={DDSL: Deep Differentiable Simplex Layer for Learning Geometric Signals},
author={Chiyu "Max" Jiang and Dana Lynn Ona Lansigan and Philip Marcus and Matthias Nießner},
year={2019},
eprint={1901.11082},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
Please contact Max Jiang if you have further questions!