AppleDiseaseNet-B7: A Robust Solution for Apple Leaf Disease Classification and Model Evaluation

Apple leaf diseases, like cedar apple rust, black rot, and apple scab, pose severe obstacles to apple farming and impact crop productivity and quality. The sophisticated classification model AppleDiseaseNet-B7, known for its efficiency and outstanding performance, was created using the EfficientNetB7 architecture and is presented in this paper. The model improves its feature extraction capabilities by utilizing pretrained weights from ImageNet, which makes it a good fit for differentiating between various apple leaf diseases. AppleDiseaseNet-B7 incorporates numerous unique layers intended to increase classification precision. A Global Average Pooling layer streamlines feature maps, complicated patterns are captured by a dense layer with ReLU activation, overfitting is reduced by a dropout layer, and class prediction is accomplished by a final dense layer with softmax activation. The model's resilience is further increased using thorough data preparation, and augmentation approaches from the PlantVillage and Kaggle Apple Leaf Disease datasets. A comprehensive comparison is made between AppleDiseaseNet-B7's performance and other well-known models, including DenseNet121, ResNet50, and AlexNet. The analysis shows that AppleDiseaseNet-B7 regularly performs better than comparable models in terms of accuracy, precision, recall, and F1 score, among other criteria. This comparison demonstrates how well AppleDiseaseNet-B7 performs in classifying apple leaf diseases, providing a reliable and effective method for managing and detecting agricultural diseases. The study emphasizes how crucial sophisticated model selection is to enhancing crop health monitoring and disease control.