Early Leaf Disease Detection and Classification Using Deep Learning Techniques

Accurate and rapid classification of maize (Zea mays) foliar diseases is essential for crop health and optimal agricultural yields. In consideration of the growing demand for sustainable agriculture and food security, the timely identification and accurate diagnosis of plant diseases have become essential. Modern deep learning techniques provide exciting opportunities for automated disease classification, which may greatly help farmers and agricultural professionals. The primary goal of this research is to develop an effective deep neural network model that accurately categorizes various diseases affecting maize leaves through the use of image data. We employed four different models, namely Simple ResNet-50, Binary ResNet-50, Otsu Threshold, and U<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> Net and CNN with segmentation, utilising CNNs to improve the precision of classification. The dataset contained images of maize leaves showing various diseases. We used three classification techniques to analyse the images. Each of these methods aimed to effectively differentiate the affected areas in leaf images, thus enhancing the model’s performance. The classification phase consisted of training and testing CNN and ResNet-50 models multiple times (50, 100, and 150) using two different data splits (70-30 and 80-20). The evaluation of these models was based on metrics such as test accuracy, precision, recall, and F1 score.