Detection and Classification of Grape Leaf Diseases using EfficientNet Model

Viticulture or grape cultivation is a significant horticulture method all over the globe. Often, grapevines suffer from many viral, fungal, and bacterial infections that affect production quantity and quality. Farmers worldwide use traditional methods to detect and treat these diseases. However, they are time-consuming, labor intensive, and require significant expertise. The EfficientNetB0 model is implemented for automating the diagnosis of grape leaf diseases. The transfer learning (TL) approach is leveraged to train and fine-tune the EfficientNetB0 model using Kaggle grapevine disease dataset. The leaf image dataset represents four classes - black rot, esca, leaf blight, and healthy samples. The model’s performance is evaluated based on metrics like accuracy and loss. The result shows that EfficientNetB0 gave accuracies of 92.45% and 96.73% on training and validation datasets, respectively. The model achieves better results in disease classification while maintaining computational efficiency.