Generated nighttime street view image to inform perceived safety divergence between day and night in high density cities: A case study in Hong Kong
Year: 2024
Type: article
Abstract: Safety perception is widely considered a fundamental aspect of urban life which significantly influences citizens' well-being and quality of life as well as having crucial impact on the nighttime economy. However, there is a scarcity in understanding nighttime safety despite the fast-growing body of urban scene auditing research based on daytime street view imagery (SVI). To fill in the gap, this study collected ~1,000 pairwise day-and-night SVIs to train a Day-ToNight (D2N) SVI generator to effectively predict nighttime SVI based on daytime counterpart using generative adversarial network (GAN). The accuracy of fake nighttime image was evaluated with SSIM, IS and human validation. Then, an online visual survey with about 46 participants was conducted to collect their perceived safety on street scenes both day and night, of which the results become training labels for machine learning to predict perceived day/night safety perception in Hong Kong. Our results revealed significant discrepancies in day and night safety perception in Hong Kong, in which micro-level streetscape features play influential roles. Second, the sky and building view in SVI matter to the accuracy of nighttime SVI generating. Third, pavements, roads, and building views significantly relate to daytime and nighttime perceived safety, as stronger safety perception is consistently found in areas with higher building density regardless day or night, whereas untended trees and grass lower the perceived safety at night. The findings provide a new perspective on understanding how streetscapes affect daytime and nighttime safety perception, providing evidence for urban design policies to high density cities like Hong Kong to facilitate nighttime attractiveness and prosperity.
Source: Journal of Urban Management
Authors: Xiaotong Ye, Yuankai Wang, Jiajing Dai, Waishan Qiu
Institutions: University of Hong Kong, University of Pennsylvania
Language: English
Subfield: Building and Construction
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,200
OpenAlex ID: W4404832062