Deep Learning Methods for Land Cover and Land Use Classification in Remote Sensing: A Review
Abebaw Alem, Shailender Kumar
Abstract
Land Cover and Land Use classification is a problem in remote sensing imagery data. The volume of remote sensing data in the earth observation, which need analysis and interpretations, is extremely increasing in this `Big Data' era. To overcome this critical challenge, Deep learning methods have been identified as a recent powerful modelling technique to extract hidden information from big remote sensing imagery for LCLU classification. Classification of RS data in earth domain is vital for land administration and environmental protection in planning and decision making. The applications of deep learning methods for LCLU classification using remote sensing imagery data become increasing in recent. Deep learning methods such as CNN, GAN and RNN are able to be used to classify remote sensing image data. We are planning to use deep CNNs models to analyse and compare their results using various parameters and three different scale remote sensing data sets: UC Merced, NWPU-RESISC45 and EuroSAT: Sentile 2.