抄録
When local governments engage in tourism promotion, they can benefit from increasing open image data. However, they face difficulties in searching for appropriate images from a large number of directories and images. This study aims to solve this issue by developing and introducing an application that facilitates the search of potential candidates for open image data. First, to display images in order of attractiveness, we developed a model to assess the attractiveness of images and assigned attractiveness scores to all images of a local government. Second, to support keyword searches with common nouns, descriptions were added to them using an existing generative model. Third, to display a list of similar images, we grouped them based on the cosine similarity of their vector representations. Then, an application was developed that could reference these metadata. This application was introduced into the actual workflows in Hida City. where it received high ratings in a user survey. This application enables public officers to search for attractive images more efficiently and comprehensively. Meanwhile, through this introductory experiment, we have found that some local governments are interested in releasing thousands of open image data. Therefore, it is considered essential to develop an integrated method.
