Web-Based Classification Optimization of Grape Leaf Diseases Using Transfer Learning in CNN to Improve Model Accuracy and Efficiency.

Abstract

Grapes are a high-value crop, but their yield is easily reduced due to leaf disease. This study focuses on developing a web-based grape leaf disease classification system that works automatically and in real time. The approach used is Convolutional Neural Network (CNN) with transfer learning using the EfficientNetB0 architecture. The research data consists of 8,000 grape leaf images divided into four classes (healthy, black rot, black spot, and leaf spot) with a composition of 80% for training, 10% for validation, and 10% for testing. The initial CNN model achieved an accuracy of 97%, which then increased to 99% after optimization using EfficientNetB0 with fine-tuning. The implementation of the system through Flask showed fast and accurate prediction results, proving that transfer learning plays an important role in improving classification performance.