Dong, J, Feng, X, Wang, L, Liu, F, Genc, B & Wu, Y 2024, 'Application of artificial intelligence recognition model methods in the analysis characteristics of closed/abandoned mine resources', in AB Fourie, M Tibbett & G Boggs (eds), Mine Closure 2024: Proceedings of the 17th International Conference on Mine Closure, Australian Centre for Geomechanics, Perth, pp. 973-990, https://doi.org/10.36487/ACG_repo/2415_70
(https://papers.acg.uwa.edu.au/p/2415_70_Xiaotong/)
Abstract: With the continuous development of China's social economy and exploitation of coal resources, some mines have reached the end of their life cycle. The precise intelligent recognition of the types, boundaries, and scope of closed/abandoned mines is a fundamental issue for energy resources, low carbon development and ecological protection. The study constructed a method for real-time live automated identification of spatial characteristics of closed/abandoned mines to obtain high-precision and high-quality information. The research included: 
Integrating multi-source remote sensing data such as Google Images, GF-6, Sentinel-2 and artificial intelligence technology to establish four datasets:
coal mine sites (open pit)
coal mine sites (underground)
coal-power sites
coal chemical sites. 
The dataset covered 21 categories of samples. Configured with six cuboids for each sample type, 6 × 10 × 21 samples were created, totalling 1,260 site samples. The optimal confidence interval ranges from 80% to 86%. 
Developing a closed/abandoned mine site classification quantitative model (CSCQM) and a closed/abandoned mine site range characteristic model (CSRCM). The average accuracy of the models is 0.837.
Take the example of China's closed shaft mine Shaanxi Zhujiahe coal mine – a quantitative and precise identification of the surface resource types of closed mines was conducted. The office area is 2,375.7 m2, residential area is 5,073.8 m2, production area occupies 5,696.2 m2, and auxiliary production area occupies 9,951.6 m2. 
Based on open-source 3D  geographic information system (GIS) technology, coupled with artificial intelligence recognition models and other cutting-edge technologies such as web databases, a comprehensive GIS database platform for closed mines has been developed using a B/S architecture. This platform encompasses system architecture design, scene design, and functional design. 
The study aims to provide methodological references and practical support for quantifying spatial resources of closed mines.

Keywords: closed/abandoned mines, multi-source data, artificial intelligence model, precise intelligent recognition, resource characteristics