A deep convolutional neural network and a random forest classifier for solar photovoltaic array detection in aerial imagery

Power generation from distributed solar photovoltaic PV arrays has grown rapidly in recent years. As a result, there is interest in collecting information about the quantity, power capacity, and energy generated by such arrays; and to do so over small geo-spatial regions (e.g., counties, cities, or even smaller regions). Unfortunately, existing sources of such information are dispersed, limited in geospatial resolution, and otherwise incomplete or publically unavailable. As result, we recently proposed a new approach for collecting such distributed PV information that relies on computer algorithms to automatically detect PV arrays in high resolution aerial imagery [1], Here we build on this work by investigating two machine learning algorithms for PV array detection: a Random Forest classifier (RF) [2] and a deep convolutional neural network (CNN) [3]. We use the RF algorithm as a benchmark, or baseline, for comparison with a CNN model. The two models are developed and tested using a large collection of publicly available [4] aerial imagery, covering 135 km², and including over 2,700 manually annotated distributed PV array locations. The results indicate that the CNN substantially improves over the RF. The CNN is capable of excellent performance, detecting nearly 80% of true panels with a precision measure of 72%.