Multi-Plant Leaf Disease Detection and Classification using Convolution Neural Network Aad Image Processing

Every nation’s basic need is the agricultural products. Disease-ridden leaves have an effect on the nation’s agricultural output and financial resources. This research provides a deep learning based plant leaf disease classification and detection method. The website (Plant Village dataset) provided the photos that were used. Since tomatoes, peppers, and potatoes are the most prevalent plant species worldwide and in Iraq specifically, we have used them as examples of certain plant species in our study. There are $\mathbf{2 0 6 3 6}$ photos of plants and their illnesses in this data set. Plant leaf diseases are sorted into 15 class labels in the proposed system using a convolutional neural network (CNN). Of these classes, $\mathbf{1 2}$ are dedicated to diseases. Several various plants that were found, including three groups for healthy leaves and various microorganisms including fungus. Consequently, we achieved exceptional precision in both training and testing; specifically, we achieved $\mathbf{9 8. 2 9 \%}$ accuracy in training and $\mathbf{9 8. 0 2 9 \%}$ accuracy in testing across all utilized data sets. Pattern recognition is necessary in order to diagnose these disorders. Using remote sensing techniques to investigate multi- and hyper-spectral picture captures is a frequent approach in this situation. Techniques that use this strategy frequently use digital image processing capabilities to accomplish their objectives. There are numerous kinds of pattern recognition algorithms that provide accurate disease detection. Plant diseases are currently detected using principal component analysis and back propagation. These algorithms are acquired by neural network training supervision.