The decline in commercial districts in Seoul, South Korea, is worsening because of excessive competition among the self-employed, COVID-19, and a decrease in offline consumption. Assessment of this decline is the first step in presenting strategies for the revitalization of commercial districts. This study explored whether cognitive decline can be measured through computer vision and machine-learning technology and the objective factors that have a dominant effect on cognitive decline. First, a machine learning model (based on ResNet152) was developed to predict perceived decline scores (PDS) using a self-developed training dataset generated from street view imagery (SVI)-based website surveys (n=3,400). We then examined how objective measures and predicted PDS were empirically correlated. The results showed the following: (1) the distribution of PDS was uneven by region, such as the district north and south of the Han River, implicating the urban context in Seoul; (2) the spatial lag model reported that the PDS increased significantly with increased floating population density, decreased sales per store, decreased new permits per area, increased old building ratio, decreased commercial building ratio, and decreased greenery; and (3) physical environmental factors, such as the commercial building and old building ratios, had a dominant impact on cognitive decline. This study contributes to providing a methodology for automatic urban cognition assessment for local governments in preliminarily investigating underdeveloped areas without incurring much cost.
Speaker: Dr. Yunmi Park
Dr. Yunmi Park, PhD, AICP, is an assistant professor at Ewha Womans University and got her Ph.D from Texas A&M University. She also has six years of professional planning experience in South Korea, which helped her to be a certified planner in both Korea and the U.S. (American Institute Certified Planner). Before joining Ewha, she worked at Auburn University, AL as an assistant professor. Her specializations and research interests are in urban shrinkage and revitalization, land-use planning, and geo-spatial analytics.
