Abstract
<jats:p>Accurate landslide detection is useful for planning and managing post-disaster reconstructions. By using the deep-learning-based approach, we can detect the landslide after a natural disaster by using satellite imagery. Currently, visual interpretation is still the most widely adopted technique for landslides mapping, which is time-consuming and costly. In the existing system, hazard and risk mapping are used to know whether the area is a hazard-prone area or not by analyzing the risks from targeted studies from previous years. To know about risk analysis, we have to analyse the occurrences of landslides, places of landslides, and the impact of dangerous events to map to the current and give predictions of the future. The proposed system collects data from satellite imagery. After collecting data, it classifies the data as sliding and not sliding for training. Then the authors do training and test the model. Later they check the effectiveness of the model.</jats:p>