chenchao15 / GridPull

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

GridPull: Towards Scalability in Learning Implicit Representations from 3D Point Clouds

Chao Chen · Yu-Shen Liu · Zhizhong Han

ICCV 2023

Citation

If you find our code or paper useful, please consider citing

@inproceedings{chao2023gridpull,
    title = {GridPull: Towards Scalability in Learning Implicit Representations from 3D Point Clouds},
    author = {Chao Chen and Yu-Shen Liu and Zhizhong Han},
    booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV)},
    year = {2023}
}

Overview

We propose GridPull to speed up the learning of implicit function from large scale point clouds. GridPull does not require learned priors or point normal, and directly infers a distance field from a point cloud without using any neural components. We infer the distance field on grids near the surface, which reduces the number of grids we need to infer. Our loss function encourages continuous distances and consistent gradients in the field, which makes up the lack of continuousness brought by neural networks.

Demo Results

ShapeNet

FAMOUS

SRB

SceneNet

SceneNet

3DFront

mattport3d

Installation

Coming soon

Dataset

Train

Test

About


Languages

Language:Python 98.3%Language:Shell 1.7%