The Coherence and Divergence Between the Objective and Subjective Measurement of Street Perceptions for Shanghai

Abstract

Recent development in Street View Imagery (SVI), Computer Vision (CV) and Machine Learning (ML) has allowed scholars to quantitatively measure human perceived street characteristics and perceptions at an unprecedented scale. Prior research has measured street perceptions either objectively or subjectively. However, there is little agreement on measuring these concepts. Fewer studies have systematically investigated the coherence and divergence between objective and subjective measurements of perceptions. Large divergence between the two measurements over the same perception can lead to different and even opposite spatial implications. Furthermore, what street environment features can cause the discrepancies between objectively and subjectively measured perceptions remain unexplained. To fill the gap, five pairwise (subjectively vs objectively measured) perceptions (i.e., complexity, enclosure, greenness, imageability, and walkability) are quantified based on Street View Imagery (SVI) and compared their overlap and disparity both statistically and through spatial mapping. With further insights on what features can explain the differences in each pairwise perceptions, and urban-scale mapping of street scene perceptions, this research provides valuable guidance on the future improvement of models.