Reconocimiento por Deep Learning de zonas con presencia de minería Ilegal aurífera en el departamento de Antioquia, Colombia

Abstract

Illegal mining is the unregulated practice of the extractive activity of a mineral in the territory; For this document, illegal gold mining in Colombia is taken into account; an activity that threatens the country's biological resources, especially the forest ecosystem, causing deforestation in the 1990s and 2000s of 310,349 hectares per year; it is estimated that at least 6,200,000 hectares of forest have been lost in recent years, which is equivalent to 5.4% of the surface of Colombia (Office for the Coordination of Humanitarian Affairs -OCHA- (2017), it is also considered a threat to other ecosystems, the water resource, and biodiversity. As of the 2010s, illegal gold mining in Colombian territory has often been linked as a source of financing for armed groups in the country and can be found in 25 of the 32 departments, concentrating on Valle del Cauca, Antioquia, Nariño, and Norte de Santander (Office for the Coordination of Humanitarian Affairs -OCHA- 2017). The development of these extractive activities by armed groups has meant that recognition of these areas, as well as early warnings to prevent deforestation, are carried out using methods other than identification in the field, given the security conditions and access to them. . To achieve the above, it is intended to use automatic detection using the supervised classification of images as input, where this type of unregulated activity occurs, for the implementation of a Deep Learning model. Using ArcGIS Pro software, which would speed up the process of detecting illegal mining in other areas of the country. Avoiding risks in the field for the technical professional, as well as early warning of deforestation due to the threat that this activity represents.