Ensuring crop health and increasing agricultural productivity depend on the diagnosis of leaf diseases.  This research work offers a sophisticated deep learning based strategy for automatically classifying leaf diseases.  To improve accuracy, the proposed method combines features from both classification and feature extraction.  Research findings indicate that the model outperforms traditional models like CNN, AlexNet, and MobileNet V2 with a high classification accuracy of 95%.  The confusion matrix and ROC curve demonstrate little misclassification and good class separation, confirming the model’s efficacy. Additionally, high precision, recall, and F1 scores across multiple disease categories confirm balanced performance. Training and validation trends indicate effective learning, with minor overfitting be addressed using dropout regularization and data augmentation. The findings establish proposed model as a trustworthy and efficient solution for automated leaf disease diagnosis in precision agriculture.