International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 05 | May 2024
www.irjet.net
p-ISSN: 2395-0072
Classification of Plant Species from Microscopic Plant Cell Images Using Machine Learning Methods Celalettin Arslan1, Volkan KAYA2 1 Master's Student, Graduate School of Natural and Applied Sciences, Department of Artificial Intelligence and
Robotics, Erzincan Binali Yıldırım University, Erzincan, Türkiye
2Associate Professor, Faculty of Engineering and Architecture, Department of Computer Engineering, Erzincan
Binali Yıldırım University, Erzincan, Türkiye ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Understanding plant biology and classification of
momentum with the rapid development of artificial intelligence [5]. These studies aim to increase accuracy rates by replacing humans in various fields. However, in a complex field such as plant classification, the integration of artificial intelligence is of great importance in order to obtain more consistent results that are not user-dependent. Artificial intelligence-supported systems can provide more reliable and consistent results by minimizing human error. Artificial intelligence can be used as an analysis method to automatically measure plant traits to aid genetic discoveries [6]. When studies on plant species classification in the literature are examined, plant leaf images are generally used as the dataset. Many image classification studies carried out to classify plants for various purposes show that classification is achieved with high success rates if appropriate data is used.
plant species stands out as an important issue in the field of biology. In recent years, with advancements in the field of artificial intelligence, the use of artificial intelligence for plant classification has increased. In this context, plant leaf images have begun to be examined with artificial intelligence. However, in the classification of plant species using artificial intelligence, the use of cell images may provide more accurate and reliable results compared to leaf images. Cell images allow for a closer focus on the genetic structure and fundamental characteristics of the plant, whereas leaf images may be more sensitive to environmental variability. Therefore, in plant classification using artificial intelligence, analyses based on cell images are preferred. In this study, microscopic cell images of four different plant species (Ficus Benjamin, Spathiphyllum, Ficus Elastica and Anthurium) were classified using machine learning methods such as KNN, SVM, Logistic Regression, Decision Trees and Random Forest. In order to classify plant species, a new data set consisting of microscopic cell images of four different plant species was created. Using this data set, plant species were classified with five different machine learning methods and their success accuracies were compared. As a result of the comparison, the best plant species classification was obtained by Random Forest with a success rate of 96.74%, and the worst plant species classification was obtained by the KNN method with a success rate of 86.05%. According to the results obtained, it was seen that microscopic plant cell images were successfully classified using machine learning methods.)
2. LITERATURE SURVEY Wu et al. present a neural network approach to recognize plant leaves in their study. The computer can automatically classify leaf images of 32 different plant species loaded from digital cameras or scanners. For this purpose, the PNN (Probabilistic Neural Network) method, which attracts attention with its fast training process and simple structure, was preferred. 12 features were extracted, which were processed by PCA(Principal Component Analysis) to generate the input vector. Experimental results show that the algorithm can work with more than 90% accuracy. Compared to other methods, this algorithm has been determined to be fast, effective and easy to implement [7].
Key Words: Machine Learning, KNN, SVM, Logistic Regression, Decision Trees, Random Forest, Plant Classification
In Doğan and Türkoğlu's research, deep learning methods were used to classify plant leaves and the performance of these methods was evaluated. They used a total of 7628 leaf images from 32 plant classes in their study. Using deep learning algorithms such as GoogleNet, AlexNet, ResNet50, Vgg16, Vgg19, they achieved successful results in the range of 97.77%-99.72% in classifying plant leaves [8].
1.INTRODUCTION Nowadays, the use of artificial intelligence is increasing. One of the areas where artificial intelligence is most used is image classification studies [1][2][3][4]. During the image classification process, artificial intelligence applications can detect qualities in images that are difficult to detect by the human eye. Thanks to these detections, the machine trains itself and decides which class the new images should belong to, with the help of different features in different image classes. Image classification studies have gained significant
© 2024, IRJET
|
Impact Factor value: 8.226
In their study, Yaman and Tuncer aimed to detect leaf diseases using deep learning and feature selection methods using 726 images of walnut leaves. Images were divided into two classes: healthy and diseased, 17 different deep learning models were evaluated and DarkNet53 and ResNet101 were
|
ISO 9001:2008 Certified Journal
|
Page 853