Abstract

ZHANG, Y.; ZAHID, M.A.H.  and  MOODLEY, T.. Elevating safety and efficiency in mining with Vision AI: From object detection to large language model-driven decision intelligence. J. S. Afr. Inst. Min. Metall. [online]. 2026, vol.126, n.2, pp.135-140. ISSN 2411-9717.  https://doi.org/10.17159/2411-9717/949/2026.

Mining operations are under growing pressure to improve safety and efficiency while dealing with aging infrastructure, complex processes, and workforce constraints. Although many sites are equipped with surveillance cameras and control systems, critical events often go unmonitored or under-analysed due to the lack of intelligent interpretation tools. Cameras typically act as passive recorders, requiring manual review by control room operators; a process that is labour-intensive, error-prone, and reactive. Vision AI is emerging as a transformative solution, combining computer vision and artificial intelligence to deliver real-time, actionable insights. This technology has evolved along two key phases: traditional object detection, and more recently, multimodal large language model integration. This paper presents solution architectures, deployment results, and key insights from real-world implementations across underground operations, open-pit truck-shovel operations, and smelter operations, demonstrating how Vision AI is reshaping mining operations to become safer, more efficient, and more intelligent.

Keywords : Vision AI; computer vision; large language models; mining safety; operational efficiency; object detection.

        · text in English     · English ( pdf )