PROPOSING EFFICIENT DEEP LEARNING SEGMENTATION MODELS FOR SOLAR PANEL DETECTION

A simple and inexpensive method for identifying rooftop and ground-mounted solar arrays over an area is to segment satellite photos. Locating solar panels in an image is the first step in estimating energy output from decentralized solar arrays connected to a traditional electric grid. Lightweight computational approaches are required for use in segmentation models for tiny devices. While state-of-the-art deep learning segmentation models provide impressive results, they are not ideal for devices with constrained computing resources because to their lengthy training timeframes, high number of floating-point operations (FLOPS), and tens of millions of parameters