USING DEEP LEARNING TECHNIQUES EXTRACTION OF STEEP SLOPE WARNING AREAS

Abstract

 The Ministry of Land, Infrastructure, Transport and Tourism has determined that there is a risk of damage to many residents on steep slopes with a slope of 30 degrees or more that are likely to collapse, and that there is a need for certain action restrictions to prevent the risk of facilitating or inducing collapse. Areas of land with a steep slope are designated as steep slope warning areas.Since it takes a huge amount of labor and cost to conduct a basic survey for designation, efficiency and automation are expected.In this research, cost reduction is expected. The purpose of this study is to extract steep slope warning areas using deep learning techniques for the purpose of improving work efficiency. A dropout layer was added to the model to prevent overfitting. For data divided into arbitrary image sizes, standardization preprocessing was added, and the accuracy improved to 80%.