Akademik Veri Yönetim Sistemi
16th INTERNATIONAL CONGRESS ON ENGINEERING, ARCHITECTURE AND DESIGN, İstanbul, Türkiye, 20 - 22 Aralık 2025, ss.1313-1320, (Tam Metin Bildiri)
Nutrition is a fundamental requirement for all living organisms and is defined as one of the most basic physiological needs in Maslow’s hierarchy of needs. Among the food products consumed by humans, grapes attract attention with their high nutritional value and strong antioxidant components. Similar to other plants, grapevines are susceptible to various diseases caused by environmental and biological factors. Therefore, the early detection of such diseases is of great importance, as it enables timely interventions and helps preserve plant health. In this study, the early-stage identification of diseases was targeted by analyzing images of grape leaves. The aim was not only to detect existing diseases but also to distinguish between different disease types through the learning ability of the model. In this context, deep learning-based Convolutional Neural Network (CNN) architectures were employed. A dataset consisting of grape leaf images was processed in the Python-based Google Colab environment, and disease detection and classification were carried out using the AlexNet, DenseNet121, EfficientNetB3, MobileNetV2, and ResNet50 architectures.All models were trained for 20 epochs, and performance metrics such as accuracy, precision, recall, and F1-score were calculated. According to the findings, the accuracy rates of the AlexNet, DenseNet121, EfficientNetB3, MobileNetV2, and ResNet50 models were 56%, 98%, 43%, 97%, and 51%, respectively. Based on these results, it was concluded that the DenseNet121 and MobileNetV2 architectures achieved the most successful performance in detecting and classifying grape leaf diseases.