Ore particles segmentation using deep learning methods

Abstract

Abstract The feed size distribution of AG/SAG mills has a big impact on the efficiency of the whole beneficiation process. In this situation is very useful instrument could be a system for estimate size distribution of ore. In this paper method for ore segmentation was developed based on deep learning algorithm and digital image processing. First U-net network was trained on manually labelled images. Then trained network used for extracting particles area and watershed used for obtain final segmentation.