Inpainting U-Net for seamless pedestrian-level wind prediction across urban morphologies

Researchers have developed a two-stage U-Net framework that models pedestrian-level wind patterns across diverse urban layouts. By using this AI-based inpainting technique, designers can predict wind flow significantly faster than traditional high-fidelity simulation methods.