Human-perceived vs actual built environment: Using human-centred GeoAI and street view images to support urban planning in Australia.

Journal: Journal of environmental management
Published Date:

Abstract

In alignment with the United Nations Sustainable Development Goals, the pursuit of safe and sustainable cities that promote well-being across all age groups has become a core objective in urban planning and environmental management. The built environment profoundly influences individuals' thoughts, emotions, and behaviors, underscoring the importance of understanding how people perceive their surroundings in relation to objectively measured urban form. Despite growing recognition of the role of perception in shaping urban experience, empirical assessments of the alignment between perceived and actual built environments remain limited. This study explores the relationship between subjective (human-perceived) and objective (measured) characteristics of the built environment through a pilot study conducted in the Melbourne metropolitan area. Using street-level imagery from the Mapillary platform and deep learning techniques, we quantify perceived built environment characteristics across the "5D" urban design dimensions: Density, Diversity, Design, Distance to transit, and Destination accessibility. These perceptions are then compared to objective spatial metrics to assess their alignment. Our analysis reveals that neighborhoods featuring compact urban form, high density, mixed land use, convenient service access, and abundant green space tend to be perceived as more livable and aesthetically appealing. However, this relationship weakens when density surpasses a threshold, leading to perceptions of overcrowding and reduced neighbourhood quality. The findings offer actionable, place-based evidence for urban planners and policymakers seeking to integrate human perceptions into planning frameworks. Additionally, the study's scalable methodology provides a foundation for developing a national database of perceived built environments in Australia, supporting broader applications in health, social equity, and environmental research.

Authors

  • Siqin Wang
    Graduate School of Interdisciplinary Information Studies, University of Tokyo, Tokyo, Japan.
  • Wenhui Cai
    School of Science, RMIT University, Melbourne, Victoria, Australia. Electronic address: S3962618@student.rmit.edu.au.
  • Qian Chayn Sun
    School of Science, RMIT University, Melbourne, Victoria, Australia. Electronic address: chayn.sun@rmit.edu.au.
  • Connor Y H Wu
    Department of Management Science and Information Systems, Oklahoma State University, Stillwater, OK, USA. Electronic address: yuhaowu@okstate.edu.
  • Xiao Huang
    Department of Anesthesiology, Beijing Chao-Yang Hospital, Capital Medical University, No. 8 Workers' Stadium South Road, Beijing 100020, Chaoyang Distinct, China.
  • Ioannis Giannopoulos
    Department of Geodesy and Geoinformation, TU Wien, Austria. Electronic address: igiannopoulos@geo.tuwien.ac.at.
  • Negar Alinaghi
    Department of Geodesy and Geoinformation, TU Wien, Austria. Electronic address: negar.alinaghi@geo.tuwien.ac.at.
  • Zhihang Liu
    School of Urban Planning & Design, Peking University Shenzhen Graduate School, Shenzhen, Guangdong, China. Electronic address: zhihang.liu@tum.de.