aw14saurabh / pointnet

PointNet: Deep Learning on Point Sets for 3D Classification

Installation

Linux

Install poetry:

curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | python -

Install pyenv and install python 3.7.9

sudo apt install pyenv
pyenv install 3.7.9

Inside the repo folder install the python environment

pyenv local 3.7.9
poetry install

Windows PowerShell

Install poetry:

(Invoke-WebRequest -Uri https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py -UseBasicParsing).Content | python -

Download and install Python 3.7.9 from the Official Site

Inside the repo folder install the python environment

poetry install

UB CCR

There is no preconfiguration or installation needed. Just copy the folder into CCR using scp

scp -r pointnet {username}@transfer.ccr.buffalo.edu:

Usage

Linux/Windows

Activate the environment

poetry shell

To train a model to classify point clouds sampled from 3D shapes:

python train.py

Log files and network parameters will be saved to log folder in default. Point clouds of ModelNet40 models in HDF5 files are already present in the data folder. Each point cloud contains 2048 points uniformly sampled from a shape surface. Each cloud is zero-mean and normalized into an unit sphere. There are also text files in data/modelnet40_ply_hdf5_2048 specifying the ids of shapes in h5 files.

To see HELP for the training script:

python train.py -h

We can use TensorBoard to view the network architecture and monitor the training progress.

tensorboard --logdir log

After the above training, we can evaluate the model and output some visualizations of the error cases.

python evaluate.py --visu

Point clouds that are wrongly classified will be saved to dump folder in default. We visualize the point cloud by rendering it into three-view images.

UB CCR

A slurm script is provided which runs the training. To schedule your job on CCR just run

sbatch slurm.sh

Citation

@article{qi2016pointnet,
  title={PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation},
  author={Qi, Charles R and Su, Hao and Mo, Kaichun and Guibas, Leonidas J},
  journal={arXiv preprint arXiv:1612.00593},
  year={2016}
}