AI-powered renewable-energy monitor: Tracking the UAE consensus progress from google earth

Engineering notes

Engineering notes will be added by the Power for Solar editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为光伏效率、钙钛矿太阳能电池、储能技术、太阳能热利用、BIPV、并网技术等高价值论文补充中文说明。

Original abstract

Automated detection of anthropogenic surface features from high-resolution imagery is an emerging area in Earth and environmental sciences. It supports renewable energy assessment, development planning, and the monitoring of sustainable practices. The increasing capabilities of AI and the availability of unrestricted, high-resolution imagery in public domains open new vistas for intelligent systems to remotely track infrastructure rollout. Monitoring of solar energy infrastructure is pivotal to meeting the UAE Consensus of tripling renewable energy capacity. In this study, we developed and evaluated deep learning-based object detection models to identify rooftop solar panels from high-resolution Google Earth imagery. A total of 1000 images from Trivandrum (India) and Sydney (Australia) were used to train and compare YOLOv5, YOLOv8, YOLOv11, and Faster R-CNN models. Among the evaluated models, YOLOv11 achieved the strongest overall performance, recording a precision of 0.917, a recall of 0.908, mAP@50 of 0.958, and mAP@50–95 of 0.559 in Trivandrum, and a precision of 0.756, a recall of 0.817, mAP@50 of 0.844, and mAP@50–95 of 0.677 in Sydney. The results suggest that AI-based approaches can significantly enhance large-scale mapping of solar panel installation, contributing to transparent, evidence-based energy policy, infrastructure planning, and sustainability assessment.

Links and sources

• PDF from original source

Comments