Transfer Learning-Based Deep Learning Model for Corn Leaf Disease Classification

Corn, also recognized as maize, is one of the most popular food grains in the world, accounting for 80% of the total volume of global trade of feed grains [1]. Corn leaf diseases pose a significant threat to global agricultural production, impacting crop yield, quality, and economy. Rapid and accurate detection of these diseases is crucial for reducing the impact to the agricultural production. In this study, we propose a transfer learning (TL) based model for the classification of corn leaf diseases. The proposed TL model integrates a convolutional neural network (CNN) known as AlexNet as it has been pre-trained on a large dataset containing more 1000 images [2]. The model is then trained on a dataset from PlantVillage [3] which consists of images of three types of corn leaf diseases, including common rust, gray leaf spot, northern leaf blight, and healthy corn leaves. To ensure better performance, the TL model, entitled NewNet, uses data augmentation techniques such as rotation, scaling, and flipping during training, effectively enhancing the model's ability to handle variations in image appearance and diseases. Transfer learning was achieved by fine-tuning the pre-trained AlexNet CNN and leveraging its learned features. Experimental results show that the TL model achieves a 99.2% accuracy for classifying common rust disease, 93.5% for gray leaf spot, 100% for healthy corn leaves, and 90.0% for northern leaf blight. Therefore, by using the TL technique, we can improve the accuracy of detecting plant diseases, specifically corn leaf diseases.