Evaluating Urban Perception: Comparing Place Pulse 2.0 Dataset Results with Images of Varied Field of View

Understanding urban perception is crucial for designing cities that en-hance human well-being. To address limited urban perception data, recent studies use large, crowdsourced datasets like Place Pulse 2.0 (PP2) for machine learning predictions. However, the accuracy of these datasets in representing real human perception is rarely examined. This study analyzes the representativeness of the PP2 dataset from a hu-man field of view (FOV) perspective. Focusing on a 400-meter segment of Spadolini Street in Milan, we compare perceived physical features, design qualities, and six urban perceptions between street view images of PP2 FOV and human FOV. Our results reveal the differences: hu-man FOV perceives more sky, roads, and sidewalks, but fewer trees and grass. In design qualities, human FOV perceives more openness but less greenness and enclosure. Beauty, liveliness, and depression scores decrease from human FOV to PP2 view, while safety and wealth scores increase. Human FOV shows more spots with high values for beauty and liveliness and fewer for wealth compared to the PP2 view. These findings underscore the importance of considering representation from a human perspective in urban studies, suggesting that the PP2 dataset may need refinement for accurate representation. Further vali-dation and improved measurement techniques are essential to better align urban design with public perception and well-being