Publication

Abstract : Tomato leaf diseases are a serious threat to crop yields, and for successful crop management, timely and accurate detection is required. In the current work, a hybrid deep learning model implemented with the Adam optimizer and pre-trained NASNet and VGG16 models is presented for effective classification of tomato leaf diseases. Our computer vision method can be integrated into autonomous robotic farming systems because it allows for real-time disease detection. The publicly released PlantVillage dataset was used for training and validation to have a strong model able to generalize among different classes of diseases. Techniques of preprocessing and augmentation were implemented to enhance feature extraction and accuracy in classification. Through the best of VGG16 and NASNet, our ensemble model shows better performance in disease classification. Experimental results illustrate that our architecture fine-tuned with strategic optimization significantly enhances accuracy and generalization. The suggested approach allows for precision agriculture through the ability to automate disease detection in robotic systems, helps in minimizing dependence on human inspection and reducing crop loss through real-time intervention