Diagnosis of Leaf Disease Using Enhanced Convolutional Neural Network

Abstract

Despite enormous remote sensing technique available, leaf disease detection continues to be a real concern. This paper Proposes a method that uses computer vision based deep learning techniques to detect 15 categories of Leaf diseases for three types of leaves namely tomato, potato and pepper bell. For the proposed method different deep learning techniques such Modified Mobilenet, Xception, Inception_Resnet_v2 has been implemented for feature extraction, training, testing and appropriate detection of the leaf disease. Of all the above methods Enhanced mobilenet proves to be best in the detection of all 15 categories of diseases. The parameters obtained from mobilenet are more promising when compared to Xception, Inception_Resnet_v2. Using the state-of-the-art architecture considerable efficiency up to 97 percentage has be achieved.