Data+ Students Use AI to Map Solar Energy in Cape Town Amid Global Energy Crisis

Jul 24, 2025

home with solar panels on roofDurham, NC — July 24, 2025 — A team of Data+ students at Duke University are harnessing the power of artificial intelligence to help solve one of the world’s most pressing challenges: access to clean, reliable energy. As part of Duke’s Climate+ program, the student-led project, titled “Energy Transition During Energy Crisis,” is using AI and aerial imagery to identify solar energy infrastructure in Cape Town, South Africa. The goal is to better understand where solar power is being adopted—and where it’s still needed.

In Cape Town, three primary types of rooftop solar devices are commonly used: solar panels, water heaters, and pool heaters. To detect these devices from aerial imagery, the team developed two machine learning algorithms. The first is You Only Look Once (YOLO), a convolutional neural network (CNN)-based object detection model that locates solar devices without distinguishing between their specific types. The second is U-Net, a CNN-based image segmentation model that generates detailed segmentation maps by identifying the exact pixels corresponding to solar panels, water heaters, or pool heaters. Additionally, the team created post-processing code to analyze the model outputs, enabling calculation of both the area and geospatial location of each predicted solar device.

Cape Town faces persistent load shedding, the scheduled disconnection of electricity power in certain areas to prevent a complete power outage when electricity demand exceeds supply. During electrical blackouts, higher-income areas are able to utilize solar panels, while lower-income areas may not be able to access a reliable source of energy due to the upfront costs.

“Reliable data on energy access is often scarce in developing regions,” said Dr. Kyle Bradbury, Managing Director of the Energy Data Analytics Lab at Duke. “By using AI to map solar adoption, we can provide valuable insights to policymakers, researchers, and organizations working to expand clean energy.”

The project is a collaboration between the Energy Data Analytics Lab, the Sustainable Energy Transitions Initiative, and the Rhodes Information Initiative at Duke (iiD). It supports Duke’s broader Climate Commitment, which aims to advance climate solutions through education, research, and community engagement. This innovative work not only contributes to global climate goals but also empowers local communities by helping to close the energy access gap.

Learn more about this project and others at our +Programs Poster Session on July 25th from 2-3:00 p.m. at the Gross Hall Energy Hub at Duke University! For a parking pass, please email ariel.dawn@duke.edu.

Media Contact:
Ariel Dawn
Rhodes iiD Communications Specialist
ariel.dawn@duke.edu
919-684-9312

Related People

Related Projects

A team of students led by researchers in the Energy Data Analytics Lab and the Sustainable Energy Transitions Initiative will develop a method to evaluate electricity access in developing countries through machine learning techniques applied to aerial imagery data. Students will first improve the accuracy of the solar array identifying...