3 papers
cs.LG2026
Click2Poly: A VLM for vector mapping buildings and walls
Nicolas Girard, Jawher Ben Abdallah, Arno Gobbin +2
Accurate vector mapping of buildings and walls is critical for geospatial applications but remains a labor-intensive process. While recent deep learning methods have improved autom…
cs.CV2026
The P Dataset: Pixels, Points and Polygons for Multimodal Building Vectorization
Raphael Sulzer, Liuyun Duan, Nicolas Girard +1
We present the P dataset, a large-scale multimodal benchmark for building vectorization, constructed from aerial LiDAR point clouds, high-resolution aerial imagery, and vectori…
cs.CV2024
Brightearth roads: Towards fully automatic road network extraction from satellite imagery
Liuyun Duan, Willard Mapurisa, Maxime Leras +2
The modern road network topology comprises intricately designed structures that introduce complexity when automatically reconstructing road networks. While open resources like Open…