Research map / Graph learning / GNN architectures
Conv-MPN: Convolutional Message Passing Neural Network for Structured Outdoor Architecture Reconstruction
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use Conv-MPN instead of L-CNN | Outdoor architecture reconstruction · SpaceNet building test set | Region F1-score: 31.9 -> 54.2 | Helped |
| Use Conv-MPN instead of L-CNN | Outdoor architecture reconstruction · SpaceNet building test set | Edge F1-score: 59.4 -> 58.7 | Hurt |
| Use Conv-MPN instead of Nauata et al. | Outdoor architecture reconstruction · SpaceNet building test set | Region F1-score: 60.8 -> 54.2 | Hurt |
| Replace spatial feature volumes with 512-dimensional vectors (ablation) | Conv-MPN, compared with vanilla GNN · SpaceNet building test set | Region metrics | Hurt |
| Remove message passing by cutting inter-node connections (ablation) | Conv-MPN (per-edge classifier) · SpaceNet building test set | Region metrics | Hurt |
| Overwrite pooled neighbor features with zero (ablation) | Conv-MPN (zero message) · SpaceNet building test set | Region metrics | Hurt |
| Add convolutional message passing (ablation) | Conv-MPN · SpaceNet building test set | Region metrics | Helped |
| Increase convolutional message-passing iterations from two to three (ablation) | Conv-MPN · SpaceNet building test set | Overall metrics | Helped |
| Use max pooling instead of sum or mean pooling for neighbor features (ablation) | Conv-MPN · SpaceNet building test set | Overall performance | Helped |
Closest papers by meaning
- Accurate and Scalable Graph Neural Networks via Message Invariance (ICLR 2025 · similarity 0.85)
- Probabilistically Rewired Message-Passing Neural Networks (ICLR 2024 · similarity 0.84)
- EdgeDiff: Edge-aware Diffusion Network for Building Reconstruction from Point Clouds (CVPR 2025 · similarity 0.84)
- Spatio-Spectral Graph Neural Networks (NeurIPS 2024 · similarity 0.84)
- Probabilistic Graph Rewiring via Virtual Nodes (NeurIPS 2024 · similarity 0.84)
- Improving the Effective Receptive Field of Message-Passing Neural Networks (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.