Quantifying Visual Navigation in Campus Open Spaces Using a Computer Vision Model

Abstract

This study presents a framework specifically designed to measure and quantify visual experiences within academic campus environments. The framework addresses the need for quantitative methods to analyze spatial experiences, focusing on key elements of the built environment, such as visible sky, greenery, and spatial enclosure. While the framework emphasizes visual components, it does not aim to analyze broader sensory or emotional experiences. Instead, it establishes a foundation for future research to explore these dimensions comprehensively. The methodology utilizes mobile phones equipped with digital cameras and GPS sensors to capture first-person visual data while participants freely navigate through campus open spaces. Computer vision techniques, including instance segmentation and convolutional neural networks, are employed to categorize architectural and natural elements within each video frame. This process quantifies the proportional composition of visual elements such as greenery, open sky, walkways, buildings, and other structures that participants encounter. The framework is implemented as a Python model that is capable of generating quantitative outcomes. Additionally, the analysis is enhanced by integrating geographic information systems (GISs) for spatial analysis, allowing us to identify navigation and visual engagement patterns. This comprehensive methodology not only quantifies the visual attributes of spaces but also interprets their impact on the behavior and experiences of campus users. This framework offers insights into how navigation choices, visual experiences, and the types of scenes encountered on campus can be understood and analyzed. The results aim to guide urban designers in better understanding university students’ open space needs by exploring the connections between natural movement patterns and visual preferences. This research complements other qualitative approaches, providing a more comprehensive perspective on campus space utilization.