Picture Urbanism

Extensive studies have been conducted by academia on urban form related quantitative studies, most of which are at the grid or block levels. Emerging street view pictures (SVPs) provided by Internet companies like Google and Baidu enable users to virtually visit cities (on streets or even indoor) by navigating between immersive 360° panoramas. While SVPs provide convenient services for end users, they also shed light on helping researchers understand the human scale urban form which can be sensed by eye balls and touched by hands/feet. In contrast to conventional site surveys for a small place, such studies are also possible for an area with very large geographical coverage, thanks to the computation capacity of personal computers. In this talk, Dr Long will present his recent studies on using SVPs for measuring human scale urban form in terms of its geometry, quality, vibrancy, perform and their temporal dynamics.

 

More details, please see the 公众微信号 of BCL.

 

In addition, we also have some studies using pictures in the other online projects of BCL like Street Urbanism and Human-scale Urban Form. Please take a look at them. 

Download
龙瀛和周垠 2017 规划师_图片城市主义.pdf
Adobe Acrobat Document 3.9 MB

Understanding City Image Using Online Pictures for 24 Main Chinese Cities

Wide use of new data resources brings about a new way and viewpoint for city study. The paper presents city image research model based on online pictures, and establishes a framework composed by content, orientation, character, and similarity of city image. This paper uses quantitative approach to recognize city image, and conducts a positive study of 24 major cities of China.

Download
曹越皓等 2017 规划师_城市意象.pdf
Adobe Acrobat Document 3.7 MB

The Innovation of City Image Cognitive Method Based on Deep Learning

Under the background of wide application of big data, technological breakthrough of artificial intelligence and transformation and innovation of urban planning, this paper, based on the three aspects of theoretical basis, data sources, and technical methods, aims to build a new cognitive method of city image based on the combination of city image theory, big data and deep learning, which depicts the city image in depth from the three dimensions of structure, type and evaluation. This paper makes an empirical research on the city image cognition of Chongqing by using Weibo data, Microsoft computer vision and Boson NLP platform, and concludes some problems in the city image of Chongqing, such as the loss of natural landscape, the lack of human elements and the needs to be optimized of the city image structure, which prove the validity and scientificity of the method.

Download
曹越皓等 2019 中国园林_城市意象.pdf
Adobe Acrobat Document 5.5 MB