SPIC-Mo: diagnosing urban infrastructure for safe, nature-responsive child mobility, Ghaziabad, India

Source abstract

Nature-based urban design is increasingly promoted for healthier, safer and more child-responsive streets, yet ecological assessment still relies on static indicators such as canopy cover and park access, which say little about how children experience a street while walking. This study proposes SPIC-Mo (Spatial Perception Interpretation through Child Mobility), a vision-based diagnostic that reconstructs pedestrian trajectories from video and reads three behavioural signatures, hesitation, scamper and edge-hanging, as adaptive responses to traffic exposure, fragmented footways and limited shade. Detection operated on a generic person class; all reconstructed trajectories were manually annotated by two independent raters (Cohen’s κ = 0.98), confirming that 59.8% were children, 81.6% of them unaccompanied. Signatures are aggregated into hotspot maps and a composite stress surface and each is linked to a specific nature-based response: shaded refuge, ecological buffering or protected walking edges. Deployed on one informal corridor in Ghaziabad, India, the framework locates environmental vulnerabilities that static metrics miss. The reading of these signatures as environmental stress is a theoretically motivated assumption, not validated against independent ground truth, so the outputs are diagnostic hypotheses. For practice, SPIC-Mo offers a rapid, non-intrusive way to screen a street network and shortlist intervention sites before participatory child engagement.