Classification of the images (Plant-21) in the dataset created with 21 different Euphorbia Taxons with the developed AI-based hybrid model

Abstract

In the present study, a dataset consisting of 21 different plant species was introduced and classified by an artificial intelligence-based hybrid model. The dataset was created by 2 experts in the field. This study, it is aimed to eliminate the difficulties encountered in the classification of 21 different plant species and to automatically classify these species by computer-aided systems. In this paper, a hybrid model (Plant21) has been developed since the accuracy rates obtained when the classification process is performed using pre-trained deep models are low. An accuracy value of 96.56% was obtained in the Plant21 model. The fact that these plant species are in the same family and the number of classes in the study is high negatively affects the performance of the models. However, the accuracy value obtained in the developed Plant21 model is quite high. This shows that the developed Plant21 model is successful in classifying plant species. In addition, the developed dataset will be made available to researchers and will enable different studies to be carried out.