2025 Volume 81 Issue 1 Article ID: 24-00045
Recently, in addition to pedestrian-centered street spaces that facilitate the flow of automobile traffic, human-centered street spaces and local context-oriented spaces have been attracting attention. It is expected that image recognition technology based on deep learning will be applied in a variety of fields. Therefore, in this study, image recognition technology is applied to streets in urban areas in Japan to evaluate the Walkability and Lingerability of pedestrian spaces and to analyze the factors that affect these spatial performances. First, an image recognition AI model (AIHCE) was used to evaluate the pedestrian space. Walking environment, convenience and comfortability of the selected pedestrian spaces in Osaka’s downtown area were evaluated by Walkability and Lingerability metrics, and the factors that affect the evaluation results were identified. Then, the effectiveness of a phased redistribution scenario of road space was examined. Finally, we proposed a method of utilizing image recognition technology to co-create a walking space that contributes to the promotion of walkable and livable cities.
