Dynamic Urban Imagery and Emotional Perception in Shanghai's North Bund: A Deep Learning Approach

Spatial restructuring has emerged as a key trend driven by the dynamics of modern city development, influencing not only the physical environment but also residents’ emotional perceptions of their urban surroundings. This study explores the dynamic changes in urban imagery and their effects on residents’ emotional perception, using the North Bund area of Shanghai as a case study. By analyzing street view images from 2013 and 2022, this research employs semantic segmentation models to identify urban spatial elements, and applies convolutional neural networks (CNN) for emotional analysis. The study examines both direct and spillover effects of various urban elements, including roads, vegetation, buildings, and notably historical districts, on emotional dimensions such as beauty, liveliness, safety, and wealth. Results indicate significant improvements in emotional perceptions, with historical districts enhancing liveliness and wealth perception but exhibiting negative spillover effects on neighboring areas. Additionally, other urban image elements, such as vegetation and building density, were found to exert complex influences on emotional responses. The study highlights the importance of considering spatial changes and their emotional impacts in urban planning, providing valuable insights for designing more livable, human-centered cities.