Awesome

3D HOUSE WIREFRAME DATASET

Description

This project is dedicated to working with 3d house wireframe datasets stored in NPZ files. It includes a Python script for reading and visualizing 3D wireframe data using the Open3D library. The dataset and related scripts are organized to facilitate easy loading, processing, and visualization of 3D wireframes.

For more details, please visit our project homepage: https://vcc.tech/research/2024/3DWire.

Dataset Download

You can download the 3D house wireframe dataset from the following link: Download Dataset

Once downloaded, extract the zip file into the npz directory as shown in the directory structure below.

Directory Structure

3D_HOUSE_WIREFRAME_DATASET/
│
├── npz/
│   ├── 00000.npz
│   ├── 00001.npz
│   ├── ...
│   └── [more NPZ files]
│
├── read_npz.py
│
└── README.md

• npz/: This directory contains the NPZ files with the 3D house wireframe data.
• read_npz.py: A Python script for reading and visualizing the wireframe data from NPZ files.
• README.md: This file, providing an overview of the project.

Usage

Prerequisites

Make sure you have Python installed along with the necessary libraries:

• open3d
• numpy

You can install these libraries using pip:

pip install open3d numpy

Running the Script

To visualize a 3d house wireframe from an NPZ file, follow these steps:

1. Update the file_path variable in read_npz.py to point to the desired NPZ file.
2. Run the script:

   python read_npz.py
   

Script Explanation

The read_npz.py script performs the following tasks:

1. Loads the vertices and lines data from the specified NPZ file.
2. Checks if the data is correctly loaded.
3. Creates a LineSet object using Open3D.
4. Visualizes the wireframe using Open3D's visualization tools.

Acknowledgments

• Thanks to the Open3D and NumPy communities for providing the essential tools for this project.
• We appreciate the RPLAN for providing the fundamental floor plans of House.
• Thank you to Sepid Hosseini for providing the code for extracting vertices and edges of the House floor plan.
• We are grateful for the scikit-geometry Python library for providing the tools to extract roof straight skeletons.
• We appreciate vector-quantize-pytorch for offering essential tools that played a key role in our network model development.
• We thank x-transformers for providing a flexible and powerful transformer implementation that supported our project.

Citation

If you find our work useful in your research, please consider citing:

@article{3DWire24,
  title={Generating 3D House Wireframes with Semantics}, 
  author={Xueqi Ma and Yilin Liu and Wenjun Zhou and Ruowei Wang and Hui Huang},
  booktitle={ECCV},
  pages={223-240},
  year={2024}
}