Graph Attention Convolution for Point Cloud Semantic Segmentation
This is a Tensorflow implementation of GACNet for semantic segmentation on S3DIS dataset.
Installation
The code is tested on Ubuntu 16.04 with Python 2.7 and TF1.12.
For data processing, PCL is needed for neighbor points searching.
Compile Customized TF Operators
Most parts here are based on PointNet++.
The TF operators are included under tf_ops, you need to compile them first.
Modify the path of your compiler and run
cd tf_ops
sh tf_compile.sh
*Update nvcc and python path if necessary.
How to use
First, you need to prepare your own dataset with the code under the folder data_processing. Slice the input scenes into blocks and down-sampling the points into a certain number, e.g., 4096.
Here, we also calculate the geometric features in advance as it is slow to put this opteration in the traning phase.
*PCL is needed for neighbor points searching here. For a prepared dataset for S3DIS, you can download it from here.
After preparing the dataset, you can run
cd net_S3DIS
python run.py
python run_test.py
for training and testing on S3DIS. Other/Customized dataset can be done in a similar way.
Citation
If you find our work useful in your research, please consider citing:
@InProceedings{Wang2019_GACNet,
author = {Wang, Lei and Huang, Yuchun and Hou, Yaolin and Zhang, Shenman and Shan, Jie},
title = {Graph Attention Convolution for Point Cloud Semantic Segmentation},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}