AUTO‐DETECTION OF GEOLOGIC FEATURES THROUGH CITIZEN SCIENCE

Virtual citizen science is a promising method for assisting in machine learning that needs large sets of processed datasets to perform. Machine learning can then be used to help teach small unmanned aerial vehicles (UAVs/ drones) to recognize geologic features in the field which will benefit geologists in research. To explore this concept, drone images of the Mecca Hills were uploaded to a public citizen science website called Zooniverse. The webpage for the project, titled Exposed Geology, has background information on machine learning, drones, geological concepts, and the drone project as a whole. We created three task options for users to participate in the project: identification of rock outcrops in images, identification of the presence of bedding and the marking of bedding planes. Each task had tutorials that direct the participant on how to label the drone images for geologic features relevant to the section. Participants gain multiple benefits from participating in citizen science as well as our project specifically. They include an increased knowledge in geologic ideas, intrinsic reward of completing important scientific tasks, and access to working on a real research project. The processed images will help teach the drone to recognize the geologic features in the field which will improve future field research and assist geologists in furthering geological knowledge.