Nonlinear Dataset

The nonlinear dataset was created to evaluate the performance of machine learning algorithms in identifying complex nonlinear patterns. This dataset was derived by regenerating the Slovak exact cropped dataset into a new nonlinear form.

The Slovak exact cropped dataset comprises 527 images of four tree species: European beech (Fagus sylvatica L.), Sessile oak (Quercus petraea (Matt.) Liebl.), Norway spruce (Picea abies (L.) H. Karst.), and European silver fir (Abies alba Mill.). Of these, 321 images were selected to generate the nonlinear dataset, including 80 images from each of the first three species and 81 images from the last. These images were captured in Včelien, Banskobystrický, Slovakia, in June 2022.

The dataset features high-resolution images taken from a distance of approximately half a meter from the trees. Each tree is represented by 15-22 images, taken from various ground angles to provide a comprehensive view of the bark. Consecutive images maintain a minimum of 60% overlap, ensuring a robust representation of each tree's bark texture.