URBAN ECHOES:
AI-Guided Exploration of Istanbul’s Cityscape Similarities
Confronting the challenge of understanding urban complexity, the Urban Echos project in Istanbul set out to reveal the historical connections of the city with Athens, Rome, Paris, and Bursa. The goal is to discern similarities across these cities by examining their urban characteristics in depth.
The Urban Echoes project provides insights into the complex structure of urban environments and enhances the understanding of residents and urban designers. By using machine learning and data analysis, the project sheds light on the detailed nature of urban features.
This comprehensive analysis involves three key techniques: satellite images are processed using U-Net for segmentation, revealing urban textures like buildings, streets, green spaces, and water bodies. Street views are dissected using the DPT model, segmenting images into elements such as buildings, streets, sidewalks, sky, and greenery. Lastly, the ResNet model is used to classify architectural styles, identifying 52 styles. These components are compared with similar elements in other cities, using standardized scores to measure similarities.
