Awesome

MVP-PCLIP (Zero-Shot Point Cloud Anomaly Detection)

[IEEE TSMC Under Review] Towards Zero-shot Point Cloud Anomaly Detection: A Multi-View Projection Framework.

by Yuqi Cheng*, Yunkang Cao*, Guoyang Xie, Zhichao Lu, Weiming Shen,

Introduction

Detecting anomalies within point clouds is crucial for various industrial applications, but traditional unsupervised methods face challenges due to data acquisition costs, early-stage production constraints, and limited generalization across product categories. To overcome these challenges, we introduce the Multi-View Projection (MVP) framework, leveraging pre-trained Vision-Language Models (VLMs) to detect anomalies. Specifically, MVP projects point cloud data into multi-view depth images, thereby translating point cloud anomaly detection into image anomaly detection. Following zero-shot image anomaly detection methods, pre-trained VLMs are utilized to detect anomalies on these depth images. Given that pre-trained VLMs are not inherently tailored for zero-shot point cloud anomaly detection and may lack specificity, we propose the integration of learnable visual and adaptive text prompting techniques to fine-tune these VLMs, thereby enhancing their detection performance. Extensive experiments on the MVTec 3D-AD and Real3D-AD demonstrate our proposed MVP framework's superior zero-shot anomaly detection performance and the prompting techniques' effectiveness. Real-world evaluations on automotive plastic part inspection further showcase that the proposed method can also be generalized to practical unseen scenarios.

Overview of MVP-PCLIP

🛠️ Getting Started

Installation

To set up the MVP-PVLIP environment, follow one of the methods below:

• Clone this repo:

  git clone https://github.com/hustCYQ/MVP-PCLIP.git && cd MVP-PCLIP
  

• Construct the experimental environment, follow these steps:

  conda create -n PCLIP python=3.9.5 -y
  conda activate PCLIP
  pip install -r requirements.txt
  

Dataset Preparation

You can choose to download original datasets and process them according to Preprocess. You are also welcome to directly download our processed datasets. All datasets need to be placed in your DATA_ROOT.

Preprocess

MVTec3D

We remove the background and project point clouds to multi_view images.

sh process_MVTec3D.sh

Real3D

We first convert Real3D to Depth data .tiff file, and then project point clouds to multi_view images. You also can get it from BaidDu Drive

sh process_Real3D.sh

Train & Test

python run_exps.py

Main Results

1. Point-wise on MVTec 3D

2. Point-wise on Real3D

3. Object-wise both on MVTec 3D and Real3D

💘 Acknowledgements

Our work is largely inspired by the following projects. Thanks for their admiring contribution.

• VAND-APRIL-GAN
• CPMF

Citation

If you find this project helpful for your research, please consider citing the following BibTeX entry.

@inproceedings{PCLIP,
  title={Towards Zero-shot Point Cloud Anomaly Detection: A Multi-View Projection Framework},
  author={Yuqi, Cheng and Yunkang, Cao and Guoyang, Xie and Zhichao, Lu and Weiming, Shen},
  booktitle={},
  year={2024}
}