SolarFinder: Automatic Detection of Solar Photovoltaic Arrays
Project Goals and Description:
SolarFinder can automatically detect distributed solar PV arrays in a given geospatial region without any extra cost. SolarFinder first automatically fetches regular resolution satellite images within the region using publicly-available imagery APIs. Then, SolarFinder leverages multi-dimensional K-means algorithm to automatically segment solar arrays on rooftop images. Eventually, SolarFinder employs hybrid linear regression approach that integrates support vector machine (SVM) modeling with a deep convolutional neural networks (CNNs) approach to accurately identify solar PV arrays and characterize each solar deployment. Students will have opportunity to deploy and benchmark the scalable performance for the implemented SolarFinder using Apache Spark (scalable performance). The goal is to explore the potential speedups of distributed SolarFinder.
Grand Challenge: Make solar energy economical.
- Source code can be found here: https://github.com/cyber-physical-systems/SolarFinder
Dong Chen, email@example.com
Student should be familiar with basic Linux and C or Python programming and willing to learn how to program Deep learning models. Familiarity with Scikit-Learn/Keras/OpenCV is a bonus.
TIME COMMITMENT (HRS/WK)
Deep Learning Programming Satellite imagery processing
Weekly meetings to review progress and set goals. Optional attendance at group lab meetings.
PREFERRED STUDENT STATUS