Gentrification is a multidimensional and context-specific process involving transformations in neighborhoods’ physical and socio-economic fabric. However, due to limitations in data availability as well as existing modeling techniques, the measurement of the physical dimensions has long been underexplored. In recent years, the rapid growth of big data and advancements in GeoAI technologies have provided new paths to facilitate large-scale, automated, and precise analyses of gentrification. To fill the gap, this chapter explores the application of deep mapping based on computer vision algorithms to integrate physical dimensions into gentrification identification models, highlighting the advantages of GeoAI in gentrification studies. Taking Wuhan City, China, as a study case, we construct a multidimensional framework combining a deep convolutional neural network (CNN) based on Baidu Street View (BSV) with a threshold model informed by socioeconomic indicators. The integration captures both physical and social upgrade processes, mapping the spatial distribution and identifying clusters of gentrification in Wuhan between 2016 and 2019, thereby providing a nuanced understanding of gentrification across the city. The chapter concludes by discussing the challenges and future directions for incorporating deep learning approaches into gentrification research, offering new insights into integrating GeoAI technologies into gentrification studies.