Fastai and Convolutional Neural Network Based Land Cover Classification

¹ Department of Computer Engineering, Pimpri Chinchwad College of Engineering, Pune 411044, India. priya.surana0204@gmail.com
² Department of Computer Engineering, PES, Modern College of Engineering, Pune 411005, India. phulpagarbd@gmail.com
³ Department of Computer Engineering, Dr. D.Y.Patil Institute of Technology, Pune 411018 pdpatiljune@gmail.com

^* Correspondence priya.surana0204@gmail.com(P.S) ; pdpatiljune@gmail.com (P.P.)

Abstract

The primary objective of this research is to create a Deep Learning model that can accurately classify satellite images into predefined categories. To accomplish this goal, we developed an effective approach for satellite image classification that utilizes deep learning and the convolutional neural network (CNN) for feature extraction. We trained our model using a labeled dataset of satellite images provided by Planet Labs, which specializes in detecting various types of land covers. By utilizing the CNN algorithm, we were able to automatically extract features from satellite data with relatively minimal processing compared to other image classification algorithms. To develop our model, we employed the Fastai library, which enables us to quickly and effortlessly achieve state-of-the-art results in image classification tasks.