Crop disease identification using state-of-the-art deep convolutional neural networks

Deep Convolutional Neural Network (CNN) based prediction models have shown their capabilities in various problems of classification and regression on image datasets. Deep CNNs have also been used by researchers for plant disease identification. In this work, we evaluate the performance of some well-known CNN architectures VGG16, ResNet50, DenseNet121, NASNet, and MobileNet V2 for plant disease identification. Although VGG16 outperforms all the other models with an accuracy of 99.61% on PlantVillage dataset, MobileNet v2 shows a comparable performance with only 2.3 million trainable parameters as against VGG16 having 33.7 million parameters.