High-Resolution Segmentation of Photovoltaic Systems: Leveraging Geoinformation and Deep Learning for Enhanced Renewable Energy Assessment

Shifting to renewable energy sources is crucial for reducing greenhouse gas emissions and mitigating the effects of climate change. Photovoltaic (PV) systems are an essential technology for generating renewable energy. These systems show considerable variability in location and time due to their various scales, installation types, and geographic distribution. Accurately identifying and segmenting PV installations—whether rooftop or ground-mounted—is essential for assessing energy potential, monitoring system performance, and informing land use and regulatory strategies.

This study introduces a new method for segmentation of high-resolution photovoltaic (PV) systems by combining geoinformation, remote sensing data, and deep learning techniques. Unlike previous research that concentrated on specific types of PV installations, our approach allows for the simultaneous prediction of multiple categories of PV systems at a spatial resolution of 0.2 meters. By utilizing automatic labeling techniques and integrating datasets such as OpenStreetMap, we employ a state-of-the-art deep learning framework to improve the segmentation process and provide accurate spatial insights into PV deployment patterns. The proposed method achieves an overall accuracy of nearly 80 %, demonstrating its effectiveness in capturing the diverse characteristics of PV installations across various environments and instilling confidence in its reliability.

Our approach has practical applications in various areas, including assessing the spatial and temporal variability of renewable energy systems, evaluating infrastructure resilience to climate and weather extremes, and quantifying the land-use impacts associated with the expansion of renewable energy. This research aids in creating integrated scenarios for energy systems that incorporate a significant proportion of renewable resources by connecting technical, environmental, and economic aspects. The findings provide valuable tools for stakeholders involved in energy system modeling, urban planning, and policy development, ultimately advancing the goal of a sustainable and resilient energy transition.