Sharrin Manor, Arjun Devarajan, Wuming Zhang, and Jeffrey Perkins explored a lage collection of imagery data provided by the U.S. Geological Survey, with the goal of identifying solar panels using image recognition. They worked closely with the Energy Data Analytics Lab, part of the Energy Initiative at Duke.
The students coded their own proof-of-principle algorithm which identified solar panels in a small test set with over ninety percent accuracy. They also painstakingly created a ground-truthed dataset that will help train future machine-learning algorithms.
Download the executive summary (PDF).
Video: The students and their mentor talk about the project.
- Environmental Science
- Energy Systems
- Machine Learning
- Electrical Engineering
Undergraduates: Sharrin Manor, Wuming Zhang, Jeffrey Perkins, Arjun Devarajan
- Richard Newell, Director, the Energy Initiative at Duke
- Leslie Collins, Elec. and Comput. Engineering
- Timothy Johnson, Energy and the Environment
Project Mentor: Kyle Bradbury, Managing Director, Energy Data Analytics Lab
- Raghav Saboo, Masters candidate, Economics
- Jordan Malof, Ph.D. candidate, Elec. and Comput. Engineering