Monitoring and analysing human–nature interaction through sensor and vision technologies

Urban nature supports environmental sustainability, public health, recreation, and everyday urban life. Rapid advances in sensing technologies, crowdsourced digital platforms, computer vision, and artificial intelligence are transforming how human activities and environmental experiences are monitored and interpreted in cities. However, research often emphasises static environmental indicators or visitor counts, with less attention to the behavioural, experiential, and semantic dimensions of human–nature interaction. This dissertation adopts a computational, human-centred approach to studying human–nature interaction (HNI) in urban environments. It aims to: (1) identify how sensor- and vision-based technologies have been used to monitor human behaviour in natural areas and assess limitations in the literature; (2) evaluate sensing systems for monitoring device-based proxies of visitor presence and movement in urban parks; and (3) develop scalable computer-vision and multimodal-AI frameworks to classify photographed representations of HNI and examine their spatiotemporal patterns. These objectives are addressed through a scoping review, field sensing experiments, crowdsourced social-media analysis, computer vision, vision–language models, and spatial analysis. The review found a strong emphasis on visitor counts and movement trajectories, with fewer studies examining activity, interaction, or environmental experience (Chapter 2). Bluetooth Low Energy analyses showed that calibrated, low-cost systems can provide scalable indicators of visitor presence and movement under suitable site conditions, but do not directly measure people or experience (Chapter 3). Using crowdsourced photographs and multimodal AI, the dissertation classified whether nature was represented as a setting, visual subject, or interaction partner (Chapter 4). Spatiotemporal analyses showed that these photographed, AI-classified representations varied across Vancouver’s urban contexts, environmental settings, and time periods (Chapter 5). Overall, the findings support a shift from viewing urban nature solely as ecological infrastructure or static green coverage towards understanding it as an experiential system shaped through movement, perception, activity, and interaction. The dissertation advances human-centred public-life research in computational urbanism by integrating sensing and AI with questions of human–nature interaction, environmental engagement, and everyday public life.