Analysis based on Different Optimization Algorithms for Landslide Detection

Abstract

The discovery of landslides is essential process in hazard and risk studies and has acquired immense focus among scientists in upcoming years. Remote sensing is an economical solution for detecting landslides and updating classical landslide databases, but discovering landslides in remote sensing data is complex and needs enhancements. The current studies reveal that the deep models enhance the landslide mapping outcomes compared to classical machine learning models. This paper presents the comparative assessment of landslide detection using different optimization algorithms to examine and justify the efficiency of landslide detection. The deep residual network (DRN) is adapted in the process of landslide detection and evaluation will be performed with various optimization algorithms, such as Competitive swarm Optimizer (CSO), Tunicate Swarm Algorithm (TSA), Particle Swarm Optimization (PSO), Water cycle algorithm (WCA), Grey Wolf Optimizer (GWO). The WCPSO-based DRN outperformed with utmost accuracy of 0.964, sensitivity of 0.980 and specificity of 0.943.