Database for classification of Araucaria angustifolia (Bertol.) Kuntze and Hovenia dulcis Thunb. based on WoldView2 image clippings

This database was used in the work Potential of machine learning and WorldView-2 images to recognize endangered and invasive species in the Atlantic Rainforest. In this work, we compared the use of spectral and textural descriptors (Pyramid Histogram of Oriented Gradients - PHOG and Filter Edge), optimized by data mining (Genetic Algorithm - GA) submitted to two classifiers Random Forest - RF and Artificial Neural Network - ANN. Thus, we sought to explore the spectral and textural response of these species in WorldView-2 multispectral images.