Publication

Abstract : Plant disease is a persistent problem for farmers, and it is one of the most serious risks to income and food security. This initiative aims to enhance the quality and quantity of agricultural output in the nation by classifying plant leaves into sick and healthy leaf types. The smart farming system is an innovative technology that aids in the improvement of agricultural quality and quantity. Deep learning using Convolutional Neural Networks (CNN) has successfully classified various plant leaf diseases. It represents a contemporary technique that offers cost-effective disease diagnosis. CNN presents a simplified version of a much broader image. In this research, we proposed a hybrid deep learning model to detect plant diseases using mixed deep learning techniques. The UNET deep learning framework has used for disease detection and classification. In convolutional neural layer feature extraction has done and pooling layer optimized those features, finally dense layer classifies the test object. The numerous synthetic and real time plan dataset has used for evaluation. In extensive experimental analysis two machine learning and two deep learning classifiers are evaluated such as SVM, PCA, CNN and modified CNN (mCNN). The mCNN is the collaboration of VGG16 and VGG16 backbone for classification and YOLOv3 model data preprocessing. The mCNN obtains 96.80% detection and classification accuracy on heterogenous dataset which is higher than other classifiers as well as conventional classifiers.