Towards Evidence-Based Urban Perception Analytics: Integrating Eye-Tracking and AIPowered Heatmaps for Pedestrian-Centered Urban Quality Assessment

Urban design often underrepresents how pedestrians perceive and visually engage with theirsurroundings. This study introduces an evidence-based methodological framework that integratesreal-world eye-tracking data with AI-powered semantic segmentation to assess visual attention inurban streetscapes. Gaze behavior was recorded using wearable eye-trackers along pedestrianroutes in a mid-density American city and categorized into seven Area of Interest (AOI) classesrepresenting key urban elements such as roads, vegetation, buildings, vehicles, and sky. Gazebased metrics including fixation duration, visit frequency, and pupil diameter—were normalizedand weighted to compute a composite Urban Attention Index (UAI), quantifying perceptualengagement across AOI classes. A complementary Urban Quality Index (UQI) was developedbased on pixel-level AOI area distributions and weighted spatial attributes such as greenness,walkability, and enclosure, offering an objective measure of environmental quality.The relationship between UAI and UQI was analyzed to reveal mismatches between perceivedattention and spatial merit such as visually dominant yet low-quality zones, or high-quality areasthat receive limited attention. These findings were operationalized through an interactive, AIbased evaluation tool, which enables designers, planners, and researchers to simulate, visualize,and compare human attention with urban spatial quality in a systematic, data-driven manner.Unlike traditional perceptual surveys or purely spatial analyses, this framework quantifiessubconscious visual engagement and juxtaposes it with spatial quality indicators, highlightingdiscrepancies between design intentions and human experience.The proposed framework, implemented through an interactive, user-friendly platform, enableseven non-experts to generate semantic segmentations, heatmaps, and insights for any street-levelurban image. The empirical findings illustrate perceptual engagement patterns: ground-levelsurfaces, construction, and natural features attract closer attention, while sky, objects, andhuman figures tend to be peripheral. By bridging the gap between human perceptual behaviorand spatial form, this approach advances both theory and practice in urban design. It establishesperceptual salience as a measurable and actionable design variable, empowering designers anddecision-makers to create public spaces that more accurately reflect actual user experience andsupport responsive, data-informed interventions.