Verification of Applicability of Anomaly Detection Method using Aerial Images of Reservoir Environment

There are many reservoirs in Japan, and there is a need for more efficient inspection and management. Therefore, in this study, we focus on automatic inspection of the reservoir environment, mainly composed of soil and grass, and propose an automatic anomaly detection method. This study proposes a method for anomaly detection from aerial images of reservoir environments using an autoencoder. Also, it describes constructing a 3D simulation environment to verify the proposed method. Simulation experiments show the applicability of the proposed method.