A Novel Approach to Grape Leaf Disease Detection using GAN-Enhanced ResNet and EfficientNet

In this research, deep learning models are applied to discover grape leaf diseases with a high sense of accuracy and precision to help address the main challenge facing the viticulture industry. To tell apart between healthy and diseased grape leaves, a neural network using ResNet and EfficientNetB5 as well as a two-meter model combining both were used using 5779 training and 360 test images. Statistics have proven that the suggested ensemble model is the most reliable, at 98.06%, and more effective than ResNet (91.23%) and EfficientNetB5 (93.12%). With an ensemble model, both the specificity and recall were 0.99 and F1 score reached a value of 0.99, therefore the model mainly avoided both false positives and false negatives. How well the output data was learned by the model and the effectiveness of using the ensemble model were seen in the training and validation loss curves as the loss for the ensemble was close to 0.1. Based on these findings, ensemble learning explains how diseases can be recognized, which helps reduce costs and damage by responding quickly in precision agriculture.