Point2Sequence: Learning the Shape Representation of 3D Point Clouds with an Attention-based Sequence to Sequence Network
Created by Xinhai Liu, Zhizhong Han, Yu-Shen Liu, Matthias Zwicker.
If you find our work useful in your research, please consider citing:
@inproceedings{liu2019point2sequence,
title={Point2Sequence: Learning the Shape Representation of 3D Point Clouds with an Attention-based Sequence to Sequence Network},
author={Liu, Xinhai and Han, Zhizhong and Liu, Yu-Shen and Zwicker, Matthias},
booktitle={Thirty-Third AAAI Conference on Artificial Intelligence},
year={2019}
}
In Point2Sequence, we build the multi-scale areas in the local region of point sets by a sequential manner. To explore the correlation between different scale areas, a RNN-based sequence model is employed to capture the contextual information inside local regions. In addition, we also introduce an attention mechanism to highlight the importance different scale areas.
In this repository we release code our Point2Sequence classification and segmentation networks as well as a few utility scripts for training, testing and data processing.
Install TensorFlow. The code is tested under TF1.4 GPU version and Python 2.7 on Ubuntu 16.04. There are also some dependencies for a few Python libraries for data processing like cv2, h5py etc. It's highly recommended that you have access to GPUs.
Before running the code, you need to compile customized TF operators as described in PointNet++.
To train a Point2Sequence model to classify ModelNet40 shapes (using point clouds with XYZ coordinates):
python train.py
To see all optional arguments for training:
python train.py -h
In the training process, we also evaluate the performance the model.
To train a model to segment object parts for ShapeNet models:
cd part_seg
python train.py
Follow the dataset in PointNet++, you can refer to here on how to prepare your own HDF5 files for either classification or segmentation. Or you can refer to modelnet_dataset.py on how to read raw data files and prepare mini-batches from them.
Our code is released under MIT License (see LICENSE file for details).
- PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space by Qi et al. (NIPS 2017)
- Attentional ShapeContextNet for Point Cloud Recognition by Xie et al. (CVPR 2018)