An Efficient Dynamic Deep Learning Methodology for Identification of Plant Disease and it’s classification

Nowadays, estimation & identification of plant diseases (PD) pointedly shows the high impact on agricultural & food productivity. In this paper, develop the Dynamic Deep Learning Methodology-(DDLM) for Plant Disease Classification-(PDC) using the Residual Neural Network (ResNet) Architecture, enhanced with an intelligent supplement recommendation module. The procedure of present Deep Learning Methodology (ResNet) is gathering the images from the number of input sensors, create large amount of dataset that contains number of sample leaf images (both diseased & healthy) finally applying ResNet model to dataset. The model is trained (80 %) on a large dataset of plant leaf images, including healthy and diseased samples across various species. Pre-processing steps such as (R_N_A) Resizing (R), Normalization (N), and Augmentation (A) are working on development of the model to improve model generalization. Once a disease is detected, Methodology generates output including the disease name (e.g., "Tomato Late Blight") and a Recommended Supplement (e.g., "Apply Copper-Based Fungicide, Ensure Proper Drainage"). The ResNet50 model, fine-tuned using Transfer Learning (TL), achieves a classification accuracy of 97.4%, outperforming traditional CNN models. Early estimation & identification of plant diseases (PD) gives the high increases the yield of the crop. Evaluation metrics such as Confusion Matrix, Precision, Recall & F1-score validate the reliability of the model across multiple classes. By integrating accurate disease detection with actionable supplement guidance, the proposed solution empowers farmers to take immediate and informed actions, enhancing crop health and yield with supplement recommendation. When comparing with resnet50 the other methods had a less accuracy. KEYWORDS— Hybrid Machine Learning Methodology (Dynamic Deep Learning Methodology-(DDL) for Plant Disease Classification-(PDC), Transfer Learning (TL), Residual Neural Network (ResNet), Image Classification, Accuracy, Disease Detection, Precision Agriculture, Smart Farming, Transfer Learning.