Solar Panels Recognition Based on Machine Learning

Abstract

Renewable energies, sustainable practices and carbon neutrality have become important goals for countries. Solar panels are a good alternative to produce energy. Monitoring, maintenance and fault detection processes represent aspects of vital importance when making concrete decisions that affects a certain percentage of the solar farms. In this paper we present a system capable of detecting solar panels location through machine learning.The main goal is to aid solar panels farm managers to locate solar panels in real time in a real area by using a machine learning model. With the use of a camera and a drone, we will be able to fly over the solar farm and identify the panels. The YOLO (You Only Look Once) object detection model is used, training and testing the neural network with a data-set of 280 images. The neural network was capable of recognize the panels in different images and videos in which we put it to the test but getting a good precision at the end.