README.md

metadata

license: other
tags:
  - diffusion
  - point-cloud
  - airplane
  - 3D
datasets:
  - shapenet

Model Description

– Luo, Shitong and Hu, Wei – 2021

Proposed a probabilistic generative model for point clouds inspired by non-equilibrium thermodynamics, exploiting the reverse diffusion process to learn the point distribution. All models are available on the original Github repo Link. It consists of a model for airplane model generating.

Documents

• GitHub Repo
• Paper - Diffusion Probabilistic Models for 3D Point Cloud Generation

Datasets

ShapeNet is a comprehensive 3D shape dataset created for research in computer graphics, computer vision, robotics and related diciplines.

• Offical Dataset of ShapeNet
• author's training dataset
• pre-trained models

How to use

# Train an auto-encoder
python train_ae.py 

# Train a generator
python train_gen.py

# Test an auto-encoder
python test_ae.py --ckpt ./pretrained/AE_all.pt --categories all

# Test a generator
python test_gen.py --ckpt ./pretrained/GEN_airplane.pt --categories airplane

BibTeX Entry and Citation Info

@inproceedings{luo2021diffusion,
 author = {Luo, Shitong and Hu, Wei},
 title = {Diffusion Probabilistic Models for 3D Point Cloud Generation},
 booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
 month = {June},
 year = {2021}
}