A systematic review of machine learning-based remote sensing data analysis for geological and mined materials characterisation

The mining industry is undergoing a significant transformation, driven by advancements in remote sensing technology that enable the collection of large-scale data on the geological and geotechnical properties of mined materials. As the volume and complexity of data generated by advanced imaging methods continue to increase, traditional analytical techniques struggle to effectively process and interpret this information. To explore current practices and the application of machine learning in interpreting complex imaging data for mine material characterisation, a review of 92 studies from 2004 to 2024 was conducted. This review focuses on key aspects of mining operations, including exploration, extraction, and waste management. It highlights the unique challenges inherent in the mining environment—particularly the heterogeneous nature of geological and mined material samples, which can result in spurious absorption features that complicate data analysis. In addition, it discusses the challenges posed by high-dimensional data resulting from sensor capabilities, as well as the cost and time constraints associated with existing algorithms. Ultimately, the review underscores both the opportunities and limitations of current machine learning approaches in analysing geological and mined materials, emphasising the need for ongoing research to overcome these challenges and fully utilise machine learning-based remote sensing in the mining sector.