• ⚙️ Requirements
• 🔍 Example Usage
• 📈 Evaluation
• 🤝 Contributing
• 📄 License
• 📚 Citation

This project depends on Habitat-Sim and the Matterport3D (MP3D) dataset. Please follow the official installation instructions in their respective repositories to set them up properly.

Additional Python dependencies include:

• matplotlib
• numpy
• pandas
• transformers
• torch
• torch_geometric

Make sure to install these packages via pip or conda before running the project.

# Build the topological graph
python ./build_raw_graph_data.py

# Plot topological graph statistics
python ./graph_stats.py

# Train GCN models
python ./leiden_GCN_train.py
python ./leiden_GINE_train.py
python ./leiden_SAGE_train.py

# Run ablation study and generate plots
python ./ablation_study.py
python ./ablation_study_plot.py

• Topological graph visualizations can be found in the ./data/ directory.
• All experimental results are saved in the ./results/ directory.

1. Fork → Clone → Create feature branch
2. Add tests for new modules
3. Submit PR → Review → Merge

This project is MIT Licensed. See LICENSE for details.

If you find our work helpful, feel free to give us a cite:

@misc{topoLGCN,
    title = {A Topological Graph Framework for Enhanced Scene Detection and Understanding in Robotic Navigation},
    url = {https://github.com/YichengDuan/topog},
    author = {Yicheng Duan, Duo Chen},
    month = {April},
    year = {2025}
}