Classification Analysis Of Regional Characteristics Using Convolutional Neural Network On Satellite Image

Abstract

One of the developments of Machine Learning technology is Deep Learning which uses an algorithm based on mathematical concepts that work like the human brain. An example of the use of deep learning is for digital image processing. Image Processing is used to identify, classify objects quickly, precisely, and can process multiple data simultaneously. In this study, an analysis of the classification of regional characteristics will be carried out. Regional characteristics are divided into two aspects, namely water areas and land areas. The land area is divided into mountains, highlands, lowlands, and valleys. While the territorial waters include straits, bays, rivers, and lakes. Classification will be done using one of the algorithms from Deep learning used in image processing, namely Convolutional Neural Network (CNN). The CNN algorithm consists of 3 main layers, namely Convolutional Layer, Pooling Layer, and Fully Connected Layer. In this study using CNN architecture with a combination of 3 Convolutional Neural Networks and 2 Fully Connected Layers. At the stage of making a regional characteristic classification system using deep learning, there are several main process stages, namely data collection, system design, training, and testing. The processed dataset is a regional image dataset originating from the satellite.