Simulated cities

September 14, 2023 - Researchers at Osaka University have developed a way to train deep-learning models to accurately assess images of urban landscapes using computer simulation. A 3D city model is generated procedurally, and an image-to-image model generates photorealistic images from the ground truth images. The result is a dataset of realistic images similar to those of an actual city, complete with precisely generated ground-truth labels that do not require manual segmentation. This approach generates large amounts of data with an impressively low amount of effort, reducing costs associated with dataset preparation and helping to usher in a new era of deep learning-assisted urban landscaping.