In the era of Industry 4.0, conventional manual activities in mining are increasingly replaced by digital technologies to enhance efficiency and accuracy. Coal remains a highly demanded energy resource, particularly in countries dependent on coal-based fuels. To support good mining practices, it is essential to strengthen the implementation and monitoring aspects of exploration. Deep learning offers a promising solution for improving coal identification and classification, especially given the wide variation in coal ranks available in the market. This study focuses on distinguishing black claystone, coal, and coal rank using a deep learning approach. The workflow began with sample identification through macroscopic characterization—including color, streak, luster, and fracture—using 2D RGB imagery. The image-based model demonstrates strong capability in accurately recognizing rock types and coal ranks. The develoved model successfully categorized 78 images of black claystone (brown color, brown streak, dull luster). It also classified 160 samples with a calorific value of 4,650 cal/gram and brownish-black streak as Sub-Bituminous (Seam B). Additionally, 252 images with a calorific value of 6,493 cal/gram and bright appearance were identified as Bituminous (Seam A). This automated classification provides an efficient tool for generating comprehensive geological interpretations and supporting data-driven mining operations
Trianda et al. (Tue,) studied this question.