Detection of Landcover Using Convolutional Neural Network

Landcover mapping is essential for a variety of applications, including environmental monitoring, urban planning, and disaster management. We present a convolutional neural network (CNN) based technique for detecting landcover in Slovakia in this research. Our aim is to compare the results of the proposed method with the results of other studies in different countries and to ease the work of scientist and industrial workers. The detection, or classification, of landcover is done manually to this day using common methods including direct contact with the area. The research area is Žitný ostrov, a region in southwestern Slovakia. The area is composed of wetlands and natural pastures. To train and assess the CNN model, we used historical Sentinel-2 images of the researched area. The model was created in Matlab using the Deep Learning toolbox framework. There are various advantages to the proposed technique. It is a completely automated process that does not require any human participation. Secondly, it is a strong approach that can deal with the variety of landcover in Slovakia. Third, the approach is scalable and easily transferable to other places. The findings of this study indicate that CNNs are a promising tool for detecting landcover in Slovakia. The proposed approach may be utilized to increase the region’s landcover mapping efficiency and accuracy as this is usually done manually.