| Sumario: | Distinguishing and understanding various plants species is of most extreme significance for keeping up biodiversity. Leaf image classification is a pivotal errand since leaves are showing high intra class variety in the leaf features like shape, color and texture. There is a need to build a classifier to recognize the plants adequately. In this paper, we proposed a new feature selection method using Grey Wolf Optimizer (GWO) technique for plant leaf classification system. Selection of features prunes the data set by choosing a subset of appropriate features from a wide pool, avoiding problems such as over fitting, poor performance and cost of computation. Selection of features refers to the process by which a subset of specific features is chosen from a pool of features that are initially available. These selected features are given to three distinct machine learning algorithms like Naïve Bayes, Support Vector Machine (SVM), and Random Forest to classify the leaves. To assess the performance of classifiers we tested on publically accessible dataset, called Flavia. From the experimental results it is shown our proposed GWO is better than Particle Swarm Optimization (PSO) algorithm in terms of number of selected features, accuracy, precision recall, and F-measure.
|
|---|