Performance Comparison of Support Vector Machine and K-Nearest Neighbor Algorithms in Leaf Recognition System Based on Android Operating System

Plants are an important factor in conservation the ecological balance. There are thousands of plant species in the world. Due to the diversity of plant species, it is very important that plant species can be detected accurately and automatically. In the study, a mobile application developed based on server which automatically detects plant species from leaf images. Flavia and Swedish databases were used in the study. Morphological properties of the leaf and local binary pattern (LBP) algorithm were used as feature extraction method. Firebase platform was used in the study to reduce the load of the mobile device using the application and also to increase the speed of the application. In the classification, support vector machines and k-nearest neighborhood methods were used. The best accuracy in the study has found to be 86% using support vector machine algorithm.