Abstract
The street is the main public space in the city. The quality of street space might affect the frequency of using the street, thereby affect the urban vitality. Thus, many previous studies have conducted the quality measurement of street space, street health benefits etc. Among them, most researchers believe that street quality and perceived comfort is directly proportional to street greening. However, in the real urban built environment, we found that part of streets with high visual green rate do not possess good perceived comfort. Therefore, a research question that needs to be in-depth discussed is does higher street visual green rate means higher comfort level? For respond this question, Nanjing (a metropolis located in the Southeast of China) was selected as the case study in here. Afterwards, 890,000 street image data were collected through Baidu maps platform. Then, the FCN (Full Convolutional Neural Network) was adopted to identify the green factors (such as green plants and mountains) from the image data based on the street elements database. Based on the recognition results, the average visual green rate of each sample street was further calculated. Meanwhile, randomly selected volunteers were invited to fill out the online questionnaire for the verification between human perceived comfort and visual green rate of street. Based on the results, this research argued that the visual green rate cannot be considered as the only factor which impact on human perceived comfort. In some streets, the effect degree of interactive interface, street infrastructure and service are higher than the visual green rate. Therefore, the research further suggested that during the urban planning process, all urban factors should be integrated considered and designed for achieve the optimum urban environment.
