Automated Detection and Classification of Medicinal Plant Leaf Diseases using CNNs

Over the past few decades, the application of neural networks in research has grown exponentially. Neural networks are widely used for diverse classification and identification tasks, including forensics and control management systems. They have also been explored in the context of plant leaf identification. This study proposes a Convolutional Neural Network (CNN) based approach for the detection of diseases in medicinal plant leaves. Early detection of plant diseases is crucial to preventing pest infestations and minimizing their adverse effects on agriculture. Neural networks serve as powerful analytical classifiers, leveraging field images of plants to accurately identify and classify diseases, thus offering a highly efficient solution for agricultural diagnostics. Such techniques involve the identification and interpretation of huge picture databases, which are often impossible to do due to technological or financial constraints. In this work a fully automated method for the recognition and detection of disease in medicinal plants leaves using computer vision and machine learning techniques has been presented. Using ven-200 dataset the CNN model achieves accuracy of 98.57%, precision score of 93.51%, recall score of 95.67% and f1 score of 93.97% in 2500 epochs.