Document Type : Original Article

Authors

• Mehdi Sadeghzadeh ¹
• Amirhossein Hoseini Khotbeh Sara ²

¹ Associate Professor, Department of Computer Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran.

² Master's Graduate in Computer Engineering, Islamic Azad University, Science and Research Branch, Tehran, Iran.

Abstract

In this study, a web-based intelligent system for kiwifruit leaf disease diagnosis through image analysis was developed. The proposed system is capable of identifying four conditions: healthy leaves, nematode infection, Alternaria, and Phytophthora. The system was evaluated using the AgriVision-Kiwi benchmark dataset, which has been employed in previous studies on kiwifruit leaf disease detection. In the first stage, the You Only Look Once version 11 (YOLOv11) algorithm was utilized to detect and extract leaf regions. Subsequently, disease classification was performed using the MobileNet-V3 deep learning architecture. The system was implemented in Python using the FastAPI framework, while data management was handled through a MySQL database.Evaluation results obtained through 10-fold cross-validation on the AgriVision-Kiwi dataset demonstrated an overall Accuracy of 99.703%, a Macro Precision of 99.7119%, a Macro Recall of 99.703%, and a Macro F1-score of 99.703%. These findings indicate the capability of the proposed system to accurately identify healthy leaves, nematode infection, Alternaria, and Phytophthora diseases. Furthermore, the results suggest that the proposed system has strong potential for practical deployment in real-world kiwifruit orchards. In addition, the web-based nature of the system enhances accessibility for users and enables its application under field conditions without requiring specialized or complex equipment.

Keywords

• Kiwifruit Leaf Disease Detection
• Deep Learning
• Convolutional Neural Network
• Web-Based Intelligent System
• Computer Vision