Recognition of Plant Diseases by Leaf Image Classification Based on Improved AlexNet

If plants and crops suffer from pests, it impacts the country's agricultural production. Farmers or specialists usually observe naked-eye plants for disease detection and identification. But time-consuming, costly and incorrect may be this technique. Adequate and accurate findings are achieved using image processing methods. The main emphasis in this paper is on identifying plant diseases and reducing crop loss, therefore increasing the efficiency of production. Our suggested study identifies the symptoms of plants at a very early stage and uses a Deep Learning (DL) method to classify the plants' diseases based on their symptoms. This method adds to the conventional AlexNet the deconvolution layer to decrease the percentage of parameters and the cost of computation, classifying images with a fully connected layer. This research article presents a technique based on improved AlexNet for classification of plants. With a best accuracy of 96.5%, the suggested method identifies different diseases in plants with less complexity.