uprightnet-cvpr2022

Upright-Net: Learning Upright Orientation for 3D Point Cloud (CVPR2022) Code and Data

https://github.com/xufangpang/uprightnet-cvpr2022

Science Score: 31.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
    Found CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
  • DOI references
  • Academic publication links
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (8.2%) to scientific vocabulary
Last synced: 10 months ago · JSON representation ·

Repository

Upright-Net: Learning Upright Orientation for 3D Point Cloud (CVPR2022) Code and Data

Basic Info
  • Host: GitHub
  • Owner: XufangPANG
  • License: gpl-3.0
  • Language: MATLAB
  • Default Branch: main
  • Size: 43.2 MB
Statistics
  • Stars: 5
  • Watchers: 1
  • Forks: 0
  • Open Issues: 1
  • Releases: 0
Created about 3 years ago · Last pushed over 1 year ago
Metadata Files
Readme License Citation

README.md

UprightNet-CVPR2022

Upright-Net: Learning Upright Orientation for 3D Point Cloud (CVPR2022) Code and Data

This repo contains the dataset and model for the paper Upright-Net: Learning Upright Orientation for 3D Point Cloud by Xufang PANG Feng LI et al.

  • The source code has been released on Gitlab, see the link (https://gitlab.com/open-source7452523/uprightnet)

Dataset - UprightNet15

  1. We use 15 classes in dataset ModelNet40, which includes 1110 training 3D models and 370 testing models. shapename.txt includes the class name of 15 categories.

  2. pointstrain.npy / pointstest.npy are original data used in our project, with the size of 111020483 for training models and 37020483 for testing models, each 3D model includes 2048 three-dimensional points.

  3. labelstrain.npy / labelstest.npy are corresponding labels for training and testing datasets with data sizes of 11101 and 3701, which are encoded with values between 0 and 14 mapping to the 15 class names listed in shapename.txt, for example, 0 = 'bed'.

  4. pidtrain.npy / pidtest.npy are binary labels for supporting points on 3D models, with the size of 11102048/3702048。

  5. angle56.txt includes 56 points uniformly distributed points sampled on a unit half sphere. By flipping the sign of sampled points, 112 uniformly sampled points on a unit sphere can be obtained.

Code

  1. The folder utils include code that is used for reading .pny files in MATLAB, which you can download from link https://github.com/kwikteam/npy-matlab

  2. rotate.m This file provides code to rotate point clouds and generates its corresponding rotation matrix, for example, input file pointstest.npy, and output rotationtest.npy where each model is rotated 100 times to gain rotated point cloud coordinates (3700020483) and rotm_test.npy with their corresponding transformation matrix.

  3. noise.m This file provides code for adding gaussian noise with parameters mu and sigma, for example, input rotationtest.npy, it output rotationtest_noise.npy, with the size of 3700020483.

Owner

  • Login: XufangPANG
  • Kind: user

Citation (citation)

X. Pang, F. Li, N. Ding and X. Zhong, "Upright-Net: Learning Upright Orientation for 3D Point Cloud," 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA, 2022, pp. 14891-14899, doi: 10.1109/CVPR52688.2022.01449.

------------------------------------------------------------------------------------------------------

@INPROCEEDINGS{9878573,
  author={Pang, Xufang and Li, Feng and Ding, Ning and Zhong, Xiaopin},
  booktitle={2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, 
  title={Upright-Net: Learning Upright Orientation for 3D Point Cloud}, 
  year={2022},
  volume={},
  number={},
  pages={14891-14899},
  doi={10.1109/CVPR52688.2022.01449}}

GitHub Events

Total
  • Issues event: 2
  • Watch event: 3
  • Issue comment event: 4
  • Push event: 1
Last Year
  • Issues event: 2
  • Watch event: 3
  • Issue comment event: 4
  • Push event: 1