A Neural Style Transfer Network for Image Augmentation of Corn Leaves using Convoluvational Neural Netwok

We proposed a novel style transfer network using a convolutional neural network for creating augmented corn leaf images in this article. The proposed network is called the Six Convoluvational Layered Neural Style Transfer (6C-NST) Network. The proposed 6C-NST model consists of six convolutional layers and five maxpooling layers. The 6C-NST was trained on the corn leaf disease dataset to create the augmented images. There are 1346 images created by the proposed 6C-NST network. The 6C-NST network created images that are used to balance the corn leaf disease dataset. The classification performance of the augmented dataset was compared with the original dataset using pre-trained models such as AlexNet, VGG16Net, and InceptionV3Net. The classification accuracy of the augmented dataset trained by AlexNet, VGG16Net, and InceptionV3Net was 93.47, 96.27 and, 97.82 respectively. the The results show that the standard transfer learning techniques performed better on the augmented dataset than on the original dataset. The simulation results demonstrate that the proposed 6C-NST model created an augmented dataset that improved the performance of the classification techniques.