Land-Use Classification Using Convolutional Neural Networks

Abstract

Abstract: Convolutional neural networks (CNNs) have been used in several classification tasks. This study aims to evaluate the performance of CNN methods for land-use classification. CNN-based model was evaluated on aerial orthophoto data for land-use scene classification. Ground-truth data set containing 25 253 records with known land-use category were used to train the CNN model to solve a practical issue. The overall accuracy of the best model on the test data set was 94.00%. The obtained results indicated that CNN mode showed high accuracy and is suitable for land-use classification tasks.