Tomato Leaf Disease Detection Using Conditional Generative Adversarial Network

The accurate and timely detection of tomato leaf disease enhances production quality, amount, efficiency that satisfies the customer with much affordable rate. The farmer efforts for harvests through the season are not misused. The Deep Learning (DL) techniques are used to identify and detect the infected crops. The conditional Generative Adversarial Network (GAN) is proposed in this paper for tomato leaf disease detection. The PlantVillage dataset is used that covers ten classes with 16012 tomato leaf images and it is preprocessed by noise removal. Then, the conditional GAN is used for leaf disease detection. The GAN with conditional setting is applied that minimize the solution space. The super-resolved images unable to has huge reliability but much realistic and huge perceptual quality. The accuracy, sensitivity, f1score and precision are used to estimate the conditional GAN performance. The conditional GAN obtains accuracy 99.79%, sensitivity 99.35%, f1-score 99.21% and precision 99.53% when compared to existing techniques like ResNet and Compact Convolutional Neural Network (CNN).