•Aim: This study explores the potential of Ilastik, an open-source software for AI-supported image classification for rapid, reproducible evaluation of Urban Green Spaces. This quantification of different visual features existent in a variety of urban Green Spaces can be used and assessed in addition to subjective experiences by humans embedded within these environmental contexts. The outcome is crucial for the design of natural environments in city-ecosystems to enhance mental health for residents. Building on the E-Co-Flourishing framework, the combination of subjective and objective Green Space assessment promotes the development research tools that advance urban research at the intersection of urban nature health and human wellbeing.
•Importance: Providing a youth-informed research tool for developing and testing accessible interventions, we capitalise on Urban Green Space composition as a powerful tool in addressing the mental health and wellbeing, especially in young people. Integrating AI-based open source software enables a relatively fast and objective evaluation process. While existing tools have advanced environmental monitoring, significant gaps remain. For example, Google Street View offers ground-level perspective but is not designed for school grounds or parks, while addressing public spaces through participatory validation.
•Research Questions:
o RQ1: How effectively does our classification system capture perceived Green Space variation?
o RQ2a: Which visual elements do young people identify as meaningful for their wellbeing, engagement, or use of these spaces?
o RQ2b: To what extent do the AI-classified categories capture or miss these youth-identified meaningful elements?
o RQ3: How can experiential knowledge validate and enrich AI-systematised compositional Green Space data to create a rapid assessment tool for school, hospital, and community Green Spaces?
o RQ4: What are current limitations and what requirements for such a tool to support E-Co-Flourishing principles?
•Methods: We selected three Oxford Living Labs for observation and rating of varied Green Space features (public park, arboretum, hospital green).Between 6-10 young people, consisting of students (YPAG-level or equivalent) engaged in a participatory observation and rating, embodying the E-Co-Flourishing emphasis on lived experience.
For objective data, we captured 360° eye-level photographs, manually annotated subsets for training, then systematically analysed by trained AI into environmental categories (such as grass, sky, wood, leaves, built structures, flowers, etc.). Subjective data came from focus groups and movement/body mapping exercises capturing emotional and sensory engagement with the Living Lab environments.
•Data Analysis: Following Ma et al. (2025), we adapted the AI training process from forest to urban settings. We calculated proportions of each environmental category, creating a Green Space Composition Index (GSCI) for correlation with subjective ratings. Youth input validated whether our category scheme captured differences they perceived as meaningful for their wellbeing and flourishing in these spaces.
•Results: The project is ongoing and will be finished by the end of November 2025. First comparisons of Green Space category proportions (e.g., more sky vs. more trees) as well as thematic insights into lived experience (quotes) and youth feedback on whether categories captured meaningful variation provides valuable insights into how urban Green Spaces can be consciously used as an intervention for supporting mental health and wellbeing.