Performance Analysis of Genetic Algorithms and KNN Using Several Different Datasets

Abstract

This research aims to increase the accuracy of the classification of mango, corn, and potato leaf types using an approach involving feature selection with a genetic algorithm (Genetic Algorithm), classification with K-Nearest Neighbors (KNN), and image processing in the HSV color space (Hue, Saturation). , Value). The dataset used consists of more than 1500 image samples for each type of leaf, with a total of 10 tests carried out. The research process begins with processing leaf images in HSV color space to extract more representative color information. Next, a genetic algorithm is applied to select the most relevant features from the processed image. The selected features are then used as input for the KNN model in the classification process. The test results show that the proposed method can achieve a classification accuracy of 87,9%. This shows that the combination of image processing in the HSV color space, feature selection using a genetic algorithm, and classification with KNN can improve performance in recognizing leaf types. This research makes significant contributions to the field of image processing and classification and shows the potential of using genetic algorithms for feature selection in pattern recognition applications.