Using Deep Neural Networks for Predicting Diseased Cotton Plants and Leafs

Complex deep learning models have proved their success in identifying plant diseases, leaf diseases with reasonable performance. But due to the complexity in the algorithm and vanishing gradient problem, it takes lot of time and requires huge computational resources to train the model. In this paper, we analyzed the performance of the deep neural networks on predicting the fresh and diseased cotton plants and leaf. We compared three different models which include-VGG16, ResNet50 and MobileNet. In our analysis, we found that VGG16 and MobileNet models give best results on train set, validation set and test set. The models are analyzed by considering the metrics-accuracy, loss and no. of correct predictions made by the model.