Deep learning-based automatic identification of minerals in images of polished sections

Automatic identification of minerals in images of polished section is highly demanded in exploratory geology since it can significantly reduce the time spent by a human expert in the study of ores, automatically provide high quality statistics of mineral distribution of different deposits. In this work we propose a deep-learning based algorithm for automatic identification of minerals in images of polished sections and present LumenStone dataset which unites high-quality geological images of different mineral associations and provides pixel- level semantic segmentation masks.