UAV Mapping and AI Recognition for Inspection of Structural Defects in Water Resources

To increase the efficiency and accuracy of traditional hydraulic construction inspections, this study proposes an automated inspection and data management system using UAV and artificial intelligence (AI) technology. A 600-meter-long part of the river was chosen as a demonstration area, with emphasis on the use of bank condition study and deterioration identification. The approach consists of three major components. Initially, high-resolution photographs were obtained by drone once a month. Second, the YOLOv8 and Unet++ models were used to segment and detect damage in the photos. Finally, a data management platform was developed to allow for the systematic integration of picture data and the automated compilation of standardised inspection reports. The results indicate that the strategy considerably improves inspection efficiency and accuracy. The combination of uav and AI technology greatly reduces inspection time and successfully inspects a 600-meter radius of the bank. The model achieves a high IOU score and several damage detection indexes in berm segmentation, demonstrating the technical solution's practicality and application promise. This study confirms the potential use of UAV and AI technologies for hydraulic structural inspection, offering an efficient and data-driven inspection solution. Future research will focus on improving the AI model's performance, broadening the range of data samples, and supporting the complete implementation of intelligent monitoring and maintenance of water conservancy infrastructure.