Deep Learning based Model for early detection of disease in Apple Leaf and Fruit with Severity level and Recommendation for Diagnosis

In an effort to address the critical issue of disease detection on apple leaves utilizing state-of-the-art deep learning techniques, a thorough research was conducted. The study explored a wide range of methods, carefully analyzing each one’s performance and effectiveness in identifying diseases in apple fruit and leaves, and finally providing insightful information in this vital field. A broad range of deep learning models, each selected for its unique qualities and possible benefits, were explored and used during the research endeavor. This massive effort produced nothing short of extraordinary outcomes. Using cutting-edge deep learning methods, this work provides a comprehensive and exact examination of the significant issue identifying apple fruit and leaf disease. We propose a deep hybrid model with combination of CNN and LSTM to detect the severity level and recommend the diagnosis. The CNN model components extracts spatial features from the images, identifying the disease patterns, while the LSTM component captures sequential dependencies, improving classification accuracy. Additionally, CNN, Exception, InceptionResNetV2, EfficientNet, RCN, ResNet, and a suggested hybrid approach were assessed for efficacy. The proposed method was notable for its quantitative precision, recall, and F1 scores of 0.96, 0.95, and 0.95, respectively, and its quantitative accuracy of 98%. This evaluation clearly demonstrates that the recommended method, which stands out for its measured precision and recall balance, yields the greatest accuracy and overall effectiveness. It offers quantitative proof of the method’s effectiveness in precisely identifying and classifying illnesses of apple fruit and leaves.The results show the differences in performance between the different models and how the suggested method, with its measurable superiority, might revolutionize the detection of apple orchard disease.