A light weight Deep Convolutional Neural network model for plant disease identification

Identification of various plant diseases often prove to be a challenging task, mainly due to the extremely vast domain of diseases available in images of leaves and it's processing techniques. This issue can be resolved by the fast-emerging machine learning concept of Networks. This research paper proposes a rather efficient and novel neural network model can make use of modern developments to create a reliable and accurate solution. Simultaneously, the designed model reduces training time and computational expense by addressing the issue of a large number of parameters. The designed neural network model trained on the PlantVillage dataset contains 11993 images with 11 different categories of classes. The model uses depth-wise separable convolutional layers, inception modules, and global average pooling to achieve the desired results. Based on the precision, recall, and Fl measure, the model scored 97.73% accuracy. The results demonstrate the model can be used effectively for recognizing disease in various plants.