Automated Rill Erosion Detection in Tailing Dams Using UAV Imagery and Machine Learning

Tailings are residual materials and waste products generated during mining and mineral processing. In recent times, there has been a notable increase in tailing dam failures, which can be attributed to the lack of specific regulations addressing stability criteria, design standards, and comprehensive monitoring of erosion in the dam’s structure caused by intense or continuous precipitation throughout the dam’s lifecycle. As manual inspection’s time and resource costs hinder effective monitoring, the present study aims to address this issue by proposing a novel approach utilizing machine learning techniques, specifically convolutional neural network (CNN) and U-Net to automate the monitoring process. The methodology includes five main steps: collecting high-quality UAV imagery data, pre-processing of images, photogrammetry, classification by CNN, and robustness test using U-Net for rill detection. To validate the proposed methodology, a real case study was employed for data collection and rill erosion identification in the research. Experimental findings from two CNNs showcased precise identification of images highlighting rill erosions at 87.75% and 89%, respectively. Furthermore, results from three U-Net variations exhibited accuracy ranging from 85.98 to 90.17%. The proposed approach efficiently detected erosion in tailings dam structures, mitigating safety risks associated with dam failures.