Street trees play an important role in the urban environment and have positive social impacts, such as improving human health. However, they can potentially be hazardous if not properly maintained. Although periodic tree assessments are necessary in urban forest management tasks, such assessments are often neglected because of the lack of trained professionals in local cities. This study aimed to develop a comparison of image classification to identify the problems associated with trees using a limited number of training samples using image classification, object detection, and semantic segmentation.