Mining activities for minerals and metals in resource-rich countries have grown since the entire global economy growth depends on these essential materials. In the hunt for these resources, illegal mining activities have increased and are a present societal challenge in many resource-rich African countries, negatively impacting the well-being of humans and the environment. Identifying these illegal mining sites is currently done via human patrols. In this project, a method is demonstrated for automatically detecting mining sites from satellite imagery. Publicly available satellite images are analyzed with current state of the art convolutional network architectures for object detection. By examining the geographical locations of the determined mining sites to registered mining sites illegal mines can be identified for further investigation. The aim of this research is to support the realization of the UN sustainable goals (‘Life on Land’).
- Fei Fang
- Philipp Schneider
- Endong Zhu
- Wenping Wang
- run
python dedup.py
andpython xml_generator.py