Deep transfer modeling for classification of Maize Plant Leaf Disease

Currently, Deep Learning is playing an influential role for Image analysis and object classification. Maize's diseases reduce production that subsequently becomes prominent factor for the economic losses in the agricultural industry worldwide. Previously, researchers have used hand-crafted-features for image classification and detection of leaf diseases in Maize plant. Nowadays, the development of the Deep Learning has allowed researchers to drastically improve the accuracy of object identification and classification. Therefore, this paper explores AlexNet model for fast and accurate detection of leaf disease in maize plant. For validating the result, we have used PlantVillage dataset. This dataset contains two categories of maize diseases namely leaf-spot based diseases (Cercospora and Gray) and the Common rust based diseases. The former category contains 1363 images and the latter category contains 929 images. One of the biggest advantages of CNN is to automatically extract the features by processing the raw images directly. By using various iteration such as 25, 50, 75 and 100, our model has obtained an accuracy of 99.16%. In future, the proposed work can be used as a practical tool to help farmer in detecting the aforementioned diseases and protect the maize crops.

Automated summary

This paper explores the use of the deep learning model AlexNet for fast and accurate detection of leaf disease in maize plants. The results show that by utilizing convolutional neural networks, the model is able to automatically extract features and achieve high accuracy.