ATTransUNet: Semantic Segmentation Model for Building Segmentation from Aerial Image and Laser Data

Abstract

Automated extraction and segmentation of building footprints from high-resolution remote sensing imagery and airborne laser scanning (LiDAR) data is fundamental for 3D city modeling and geospatial analysis. We propose ATTransUNet, a novel architecture integrating Attention Gated Networks with Transformer self-attention modules within a U-Net backbone. The model demonstrates superior boundary delineation and robust feature fusion in the MapAI competition benchmark.

Yunus Serhat Bıçakçı

Assistant Professor

Assistant Professor specializing in GeoAI, Multimodal Vision-Language Models, and Spatial Data Science.