# Pointfilter

**Repository Path**: pdycp/Pointfilter

## Basic Information

- **Project Name**: Pointfilter
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2024-07-30
- **Last Updated**: 2024-07-30

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Pointfilter: Point Cloud Filtering via Encoder-Decoder Modeling
This is our implementation of Pointfilter, a network that automatically and robustly filters point clouds by removing noise and preserving their sharp features.

<p align="center"> <img src="Images/pipeline.png" width="75%"> </p>

The pipeline is built based on [PointNet](http://stanford.edu/~rqi/pointnet/) (a patch-based version of [PointNet](http://stanford.edu/~rqi/pointnet/)). Instead of using STN for alignment, we align the input patches by aligning their principal axes of the PCA with the Cartesian space.

## Environment
* Python 3.6
* PyTorch 1.5.0
* Windows 10 and VS 2017 (VS 2017 is used for compling the chamfer operators)
* CUDA and CuDNN (CUDA 10.1 & CuDNN 7.5)
* TensorboardX (2.0) if logging training info. 


## Datasets
You can download the training datasets used in this work from the following [link](https://entuedu-my.sharepoint.com/:f:/g/personal/n1805982j_e_ntu_edu_sg/Er5PVpfMIBZDiucsZSUX-AsB8QXXHIfzVfENWSj1u9TNng?e=wEFDZY), or prepare yourself datasets and change corresponding codes in `Pointfilter_DataLoader.py`. Create a folder named Dataset and unzip the files on it. In the datasets the input and ground truth point clouds are stored in different files with '.npy' extension. For each clean point cloud `name.npy` with normals `name_normal.npy`, there are 5 correponsing noisy models named as `name_0.0025.npy`, `name_0.005.npy`, `name_0.01.npy`, `name_0.015.npy`, and `name_0.025.npy`.  


## Setup
Install required python packages:
``` bash
pip install numpy
pip install scipy
pip install plyfile
pip install scikit-learn
pip install tensorboardX (only for training stage)
pip install torch==1.5.0+cu101 torchvision==0.6.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html
```

Clone this repository:
``` bash
git clone https://github.com/dongbo-BUAA-VR/Pointfilter.git
cd Pointfilter
```

Compile Chamfer Opertors (only for evaluation)
``` bash
cd ./Pointfilter/Customer_Module/chamfer_distance
python setup.py install
```

## Train
Use the script `train.py` to train a model in the our dataset (the re-trained model will be saved at `./Summary/Train`):
``` bash
cd Pointfilter
python train.py
```

## Test with Pre-trained Model (The filtered results will be saved at `./Dataset/Results`)
``` bash
cd Pointfilter
python test.py --eval_dir ./Summary/pre_train_model
```
## Test with Re-trained Model
``` bash
cd Pointfilter
python test.py --eval_dir ./Summary/Train
```
  
## Visual Results

<p align="center"> <img src="Images/Visual Results.png" width="100%" </p>

## Citation
If you use our work, please cite our [paper](https://arxiv.org/abs/2002.05968):
```
@article{zhang2020pointfilter,
  title={Pointfilter: Point cloud filtering via encoder-decoder modeling},
  author={Zhang, Dongbo and Lu, Xuequan and Qin, Hong and He, Ying},
  journal={IEEE Transactions on Visualization and Computer Graphics},
  year={2020}
}
```

## Acknowledgements
This code largely benefits from following repositories:
* [PointNet](http://stanford.edu/~rqi/pointnet/)
* [PCPNet](https://github.com/paulguerrero/pcpnet)