Machining tool identification utilizing temporal 3D point clouds

This repository provides the code for our paper Machining tool identification utilizing temporal 3D point clouds

Please consider citing this work if you find it influential or useful in your research.

@article{Zoumpekas2023JIMS,
author = {Zoumpekas, Thanasis and Leutgeb, Alexander and Puig, Anna and Salamó, Maria},
title = {Machining tool identification utilizing temporal 3D point clouds},
journal = {Journal of Intelligent Manufacturing},
doi = {10.1007/s10845-023-02093-5},
year = {2023}
}

Installation

Please install Python Poetry. Then simply run (inside the project directory):

poetry install
poetry shell
pip install torch==1.12.1+cu102 torchvision==0.13.1+cu102 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu102

The code was tested with CUDA 10.2 and PyTorch 1.12.1 on Ubuntu Linux 20.04.

Usage

If all the dependencies installed correctly then you can train and test new models from scratch.

Datasets

The temporal 3D point clouds are available here: (TPCOMP dataset) https://doi.org/10.34810/data205

• 16 Tools Large Dataset: Please put it under data/16_Tools_Large_Dataset/
• 16 Tools Small Dataset: Please put it under data/16_Tools_Small_Dataset/

Training

python train_classification_tools_L --model pointnet_cls --batch_size 8

Testing

python test_classification_tools_L --log_dir "path_to_trained_model_log"

Our code is mainly based on the following repository: https://github.com/yanx27/Pointnet_Pointnet2_pytorch

License

MIT License.