Publication - Article
Automatic Classification of Felsic, Mafic, and Ultramafic Rocks in Satellite Images from Palmira and La Victoria, Colombia
Date of Activity
Manually inspecting and analyzing satellite images can lead to numerous errors and is quite time consuming. Our geological contribution is to offer a means for the automatic classification of areas with felsic, mafic, and ultramafic rocks via machine learning using satellite images from Palmira and La Victoria, Colombia. Specifically, this study focuses on two types of satellite images taken from the Earth Observation System (EOS), namely natural color (bands B04 B03 B02) and infrared color vegetation (B08 B04 B03). The following machine learning algorithms were used in this study: Random Forest, K-Nearest Neighbors, Support Vector Machines, Logistic Regression, and Multilayer Perceptron. The model generated with K-Nearest Neighbors performed best for classifying natural color images with an accuracy of 91%, a precision of 87%, and a recall of 88%. Random Forest was the best model for classifying infrared images with an overall accuracy of 83%, a precision of 31%, and a recall of 31%.
Bosquez, S., Alférez, G.H., Ardila, A.M.M., Clausen, B.L. (2022). Automatic Classification of Felsic, Mafic, and Ultramafic Rocks in Satellite Images from Palmira and La Victoria, Colombia. In: Arai, K. (eds) Intelligent Computing. SAI 2022. Lecture Notes in Networks and Systems, vol 507. Springer, Cham. https://doi.org/10.1007/978-3-031-10464-0_36