Associations between Street-View Perceptions and Housing Prices: Subjective vs. Objective Measures Using Computer Vision and Machine Learning Techniques

This study investigated the extent to which subjectively and objectively measured street-level perceptions complement or conflict with each other in explaining property value. Street-scene can be assessed from self-reported survey questions, quantified land use data pixel ratios of physical features extracted street-view imagery. Prior studies mainly relied on objective indicators describe found that a better street environment is associated price premium. While very few have addressed impact subjectively-assessed perceptions. We hypothesized human subtle relationship cannot comprehensively captured indicators. Subjective measures could more effective perceptions, thus might explain housing variations. To test hypothesis, we both six pairwise eye-level (i.e., Greenness, Walkability, Safety, Imageability, Enclosure, Complexity). then their coherence divergence for perception respectively. Moreover, revealed similar opposite effects house prices Shanghai using hedonic model (HPM). Our intention was not make causal statements. Instead, set address coherent conflicting two people’s behaviors preferences. method high-throughput by extending classical urban design measurement protocols current artificial intelligence (AI) frameworks urban-scene understanding. First, percentage increases attributable were significant subjective measures. scores explained variance over scores. Second, exhibited signs Greenness Imageability results indicated simply extract recombine individual streetscape pixels fully capture For perceptual qualities familiar average person (e.g., Imageability), framework exhibits performance. Conversely, whose connotation are self-evident Greenness), outperform counterparts. demonstrates holistic understanding street-scene relations values. It also sheds light future where further stressed.