A FastAPI-based web service that processes 3D point cloud files (PTX, XYZ, or PLY format) and a YAML configuration to generate structured BIM models in IFC format. Based on the open-source Cloud2BIM project, this service segments building elements such as slabs, walls, and openings, and returns both an IFC file and a point-to-element index mapping.

• Accepts .ptx, .xyz, and .ply point cloud formats

• Automatically converts multiple point cloud files to PLY format for processing

• Uses Cloud2BIM segmentation logic

• Asynchronous processing with job tracking

• YAML-based configurable pipeline

• REST API with progress reporting

• Outputs:

  • IFC model file (.ifc)
  • JSON file mapping IFC elements to point indices

• Includes a CLI client tool

• Python 3.10+
• pip
• Optional: BlenderBIM or other IFC viewer for result validation

git clone <your-repository-url> # Replace with your actual repository URL
cd Cloud2BIM_web
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

The easiest way to run the application is using Docker Compose:

# Build and start the container
docker-compose up -d

# View logs
docker-compose logs -f

# Stop the container
docker-compose down

You can also build and run the Docker container manually:

# Build the Docker image
docker build -t cloud2bim .

# Run the container
docker run -p 8001:8001 -v $(pwd)/jobs:/app/jobs --name cloud2bim cloud2bim

The API will be accessible at http://localhost:8001/docs

# Option 1: Using Python directly
uvicorn app.main:app --reload --port 8005

# Option 2: Using the main.py script
python main.py

Visit http://localhost:8005/docs for Swagger UI when using Option 1, or http://localhost:8001/docs when using Option 2 or Docker.

python client/client.py tests/data/scan6.ptx tests/data/sample_config.yaml --server_url http://localhost:8005 --output_dir ./output

This command will:

1. Upload scan6.ptx and sample_config.yaml to the server.
2. Poll the server for job status updates.
3. Once the job is complete, download model.ifc and point_mapping.json to the ./output/ directory, prefixed with the job ID.

Each job creates a folder under jobs/<job_id>/ containing:

• input/ - Original uploaded point cloud and configuration files.
• output/ - Generated model.ifc and point_mapping.json files.
• job_status.json - Contains the current status, progress, and stage of the job.

See developer-guide.md (to be created) for full development instructions.

It's recommended to use black for code formatting and flake8 for linting.

pip install black flake8
# To format
black .
# To lint
flake8 .

• No authentication or file size limits
• No automatic cleanup of old jobs
• Limited to single-instance async workers (no Celery)

Contributions are welcome! Please open an issue or PR.

MIT License — see LICENSE file for details.

Built using Cloud2BIM and IfcOpenShell.

Happy building 🏠🚀