Abstract

Northern leaf blight (NLB) is one of the diseases responsible for significant yield loss in maize, but scouting wide areas for accurate diagnosis is time-consuming and difficult. Nowadays we applied new techniques like Artificial Intelligence for the detection of plant disease and specific as well as precise application of disease management. We show that the proposed method can reliably recognize northern leaf blight (NLB) lesions in images of maize plants collected in the field. This method employs a convolutional neural network (CNN) pipeline to solve the complexities of minimal data and the various variances that occur in images of field-grown plants. CNN's model was trained to classify NLB disease and healthy leaf. Experiments were carried out with the Efficient CNN model to classify the entire image as containing NLB and healthy leaf. Proposed lightweight CNN method achieved 96.7 per cent accuracy with minimum inference time as compared to state of the art methods. The fourth coming era has importance of AI enabled devices such as aerial or ground vehicles, will aid in automated high-throughput plant phenotyping, precision breeding for disease resistance and reduced use of fungicides through targeted application across a wide range of plant and disease categories.