LeafGuardNet: A Deep CNN-Based Model for Early Plant Disease Prediction in Smart Agriculture

The timely identification of plant diseases is very vital in crop productivity and food security. The conventional hand inspection systems are costly in time, inaccurate, and highly reliant on the level of expertise. As a solution to these issues, LeafGuardNet, which is a deep Convolutional Neural Network (CNN)-based model, will be suggested in this research study to effectively predict plant leaf diseases with high accuracy and early onset in the smart agriculture setting. The model uses a huge amount of annotated leaf image data, which is preprocessed by image enhancement and noise reduction algorithms. The proposed model uses a multi-layer CNN model, which has been optimized to extract features, dropout, and batch normalization layers to enhance accuracy and minimize overfitting. The pre-trained models, including ResNet50 and EfficientNet-B0, are incorporated in transfer learning to improve the performance of the classification. Proposed system is trained and tested against publicly available datasets like PlantVillage and field-acquired images with an overall classification accuracy exceeding 97 percent in a variety of crop species. The experimental findings reveal that LeafGuardNet is better than currently existing models in terms of precision, recall, and F1-score, which is why it is a stable decision-support tool to use in precision agriculture and sustainable crop management.