This study demonstrates how deep learning technologies can enhance the efficiency and accuracy of urban spatial layout optimization, cultural heritage preservation, and transportation network planning. The study explores the application of deep learning in the digital transformation of urban planning, specifically focusing on Beijing’s central axis, an area rich in historical and cultural significance. By utilizing an improved DeepLabv3+ model, pixel-level classification of urban functional zones was achieved, with accuracy exceeding 90%, significantly surpassing traditional models. Furthermore, the study applied an enhanced graph attention network to analyze the spatial relationships between different regions, plots, roads, and intersections, revealing the complex interactions between urban functions and connectivity. These findings validate deep learning’s capability to accurately capture intricate spatial features, optimize urban layouts, and support data-driven decision-making. The study also underscores the importance of balancing technological innovation with cultural heritage preservation and community involvement. A strategic framework for integrating artificial intelligence and big data into smart city development is proposed, offering valuable guidance for policymakers, data scientists, and engineers. This framework contributes to the advancement of sustainable and intelligent urban environments.
Anni Zhang (Fri,) studied this question.