Smartphone-based mapping of home ignition zone attributes for wildland–urban interface fire risk assessment

Abstract

Background Assessing home ignition zone attributes is essential for reducing wildfire risk in wildland–urban interface (WUI) communities; however, traditional satellite imagery often lacks sufficient spatial resolution and timely updates to capture localized fuel conditions near structures. Aims This study develops and validates a smartphone-based framework for real-time detection, mapping and 3D visualization of building features and defensible space fuel to support wildfire risk assessment and mitigation in WUI environments. Methods The smartphone-based framework was developed by training multiple YOLO models using 4413 annotated images representing 38 home ignition zone attributes. The best-performing model was integrated with multi-object tracking, monocular depth estimation and smartphone sensor data to geolocate detections and generate 3D maps from real-world videos. Key results The proposed framework successfully detected and mapped home ignition zone attributes. YOLOv11 achieved the best performance with near-real time processing capability. The integrated framework enabled approximate geospatial localization and visualization of home ignition zone attributes, capturing fine-scale fuel conditions and recent environmental changes unavailable in satellite imagery. Conclusions and implications The proposed framework provides a low-cost, scalable approach for mapping home ignition zone attributes and supports real-time fuel characterization, defensible space assessment, wildfire risk analysis, mitigation planning, emergency response and community preparedness in WUI communities.