The research work systematically investigates the use of Deep Learning (DL) approaches in smart agriculture for plant leaf disease identification. Advanced techniques such as LSTM networks, RNNs, and CNNs were employed to create robust illness detection models. Training these models on extensively annotated datasets with transfer learning and data augmentation methods significantly improved their performance. The findings indicate impressive outcomes, with DL models achieving an average recall of 96%, accuracy of 95%, precision of 94%, and F1 Score of 95%. These metrics demonstrate the models' ability to accurately distinguish between healthy and unhealthy foliage, with minimal false positives and false negatives. The results highlight the potential of DL advancements in revolutionizing plant disease detection within smart agricultural systems, contributing to improved food security and sustainable farming practices. © 2024 Elsevier B.V., All rights reserved.