Classification bee subspaces by machine learning methods

In this research, it is aimed to classify the bee species according to the intersection points on the bee wings. The intersection points on the pictures of bee wings taken from five different country / districts have been determined. The intersection point selection algorithm is proposed for the determination of intersection points on bees' wings with minimum error and for selecting a standard intersection point. Using the intersection points, 27 morphological features including angle, length and area information were extracted. The normalization process has been applied to eliminate numerical differences between these features and to reduce workload. Support Vector Machines (SVM), Artificial Neural Networks (ANN), K-Means (K-Means) and K-Nearest Neighbors (KNN) algorithms were used as the classification method. Brute Force Method (BFM), Forward Sequential Selection(SFS) and Linear Discrimant Analysis (LDA) methods were used to reduce the data size. The highest success rate in the classification of bee subspecies was 50% with K-Means, 71% with SVM, 55,3% with KNN and 82,7% with ANN. The success rates of the proposed and unused algorithms for intersection points and the success rates of cases where size reduction methods are used or not are examined. The selection of the intersection points on the images, feature extraction and the classification of data, MATLAB© program is used.