Citrus disease detection and classification using based on convolution deep neural network

ÇETİNER H.

Microprocessors and Microsystems, cilt.95, 2022 (SCI-Expanded, Scopus) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 95
  • Basım Tarihi: 2022
  • Doi Numarası: 10.1016/j.micpro.2022.104687
  • Dergi Adı: Microprocessors and Microsystems
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Communication Abstracts, Computer & Applied Sciences, INSPEC, Metadex, Civil Engineering Abstracts
  • Anahtar Kelimeler: Agriculture, Citrus diseases, Convolutional neural networks, Deep neural network, Leaf classification
  • Isparta Uygulamalı Bilimler Üniversitesi Adresli: Evet

Özet

Plant diseases that cause instability in the food supply reduce agricultural yield and production. As a result, it causes significant economic losses. Citrus, one of the plants, is widely grown all over the world and is used in many fields, especially in nutrition and health. Citrus is an agricultural product of great economic importance worldwide. However, citrus production is severely damaged by pests and various diseases. As a result, serious yield and quality losses are experienced in citrus production. In addition to the reasons stated, more than half of the products are not used in citrus production every year due to different plant diseases and environmental factors. In recent years, depending on the development of technology, image processing and machine learning algorithms have been used in many fields, including agriculture. This provides an opportunity for early detection and classification of plant diseases. In this study, it is aimed to detect and classify blackspot, canker, and greening diseases, which are frequently seen in many different regions, through images. For this purpose, first, preprocessing and segmentation processes are performed on different images from the Citrus Leaves Prepared data set in the literature. Afterward, a unique architecture based on CNN is developed. The developed architecture consists of four blocks and short paths. Each block has convolution, pooling, and batch normalization layers connected in series and parallel. From the test results of the created architecture, the average values of 95%, 96%, 95%, 96% were obtained for F1-score, Precision, Recall, and Accuracy (%) respectively. According to the findings obtained from the study, the proposed model primarily defines citrus black spot, citrus bacterial canker, and citrus disease defined as huanglongbing. It then distinguishes these diseases from each other with high accuracy. The proposed model is presented as a useful decision support tool for citrus growers to recognize and classify citrus diseases.