Research map / Vision-language and multimodal / Cross-modal learning
Urban2Vec: Incorporating Street View Imagery and POIs for Multi-Modal Urban Neighborhood Embedding
Zhecheng Wang, Haoyuan Li, Ram Rajagopal
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 Urban2Vec instead of AE+POI | Neighborhood embeddings for real estate prediction · Bay Area, average apartment sale price | R²: 0.391 -> 0.487 | Improved |
| Use Urban2Vec instead of SL+POISTATS | Neighborhood embeddings for real estate prediction · Bay Area, average apartment sale price | R²: 0.417 -> 0.487 | Improved |
| Use Urban2Vec instead of demographic attributes as inputs | Real estate prediction · Bay Area, average apartment sale price | R²: 0.445 -> 0.487 | Improved |
| Use Urban2Vec instead of AE+POI | Neighborhood embeddings for real estate prediction · Bay Area, average office sale price | R²: 0.308 -> 0.443 | Improved |
| Use Urban2Vec instead of SL+POISTATS | Neighborhood embeddings for real estate prediction · Bay Area, average office sale price | R²: 0.328 -> 0.443 | Improved |
| Use Urban2Vec instead of demographic attributes as inputs | Real estate prediction · Bay Area, average office sale price | R²: 0.337 -> 0.443 | Improved |
| Add POI word embeddings to street-view embeddings instead of POI category statistics (ablation) | SVE-based neighborhood representations · Three metropolitan areas, demographic prediction | Average R²: ~0.02 | Improved |
| Add customer review words to POI textual data (ablation) | Urban2Vec · Three metropolitan areas, demographic prediction | Average R²: ~0.02 | Improved |
| Combine street-view and POI information rather than use only one modality (ablation) | Urban neighborhood representations, excluding SL-based methods · Three metropolitan areas, demographic prediction | R² | Improved |
| Use geospatially trained SVE instead of an autoencoder image embedding (ablation) | Street-view-based neighborhood representations · Three metropolitan areas, demographic prediction | R² | Improved |
| Use POI category statistics instead of POI word embeddings (ablation) | POI-only neighborhood representations · Three metropolitan areas, demographic prediction | Prediction performance | Improved |
| Use POI word embeddings instead of POI category statistics alongside street-view embeddings (ablation) | SVE-based neighborhood representations · Three metropolitan areas, demographic prediction with SVR | Average R²: ~0.02 | Improved |
Closest papers by meaning
- Urban Region Embedding via Multi-View Contrastive Prediction (AAAI 2024 · similarity 0.88)
- GeoFidelity-Bench: Evaluating Block-Conditioned Geographic Fidelity of Street-View Generation (NeurIPS 2026 · similarity 0.87)
- Improving Region Representation Learning from Urban Imagery with Noisy Long-Caption Supervision (AAAI 2026 · similarity 0.86)
- Textual Supervision Enhances Geospatial Representations in Vision-Language Models (ICML 2026 · similarity 0.85)
- Learning Content-Enhanced Mask Transformer for Domain Generalized Urban-Scene Segmentation (AAAI 2024 · similarity 0.84)
- Geolocation Representation from Large Language Models Are Generic Enhancers for Spatio-Temporal Learning (AAAI 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.
