Tomato Leaf Disease Detection and Classification by Using Novel CNN Model

The recognition and early prevention of leaf diseases on time is very essential for improving the crop production. Different deep learning models are proposed by the researchers for identification of diseases in plants. A novel CNN model is implemented in this paper to identify and diagnose plant diseases. The standard CNN model requires a huge no. of parameters, cost of computation is high and also training time is more. In the place of the standard convolution depthwise separable convolutions are used in order to minimize the no. of parameters and computation cost. The implemented CNN architecture attained better classification accuracy of 99.05% when compared with other models like standard inception V3 98.45%, ResNet101V2 98.52%, EfficientNetB0 99.34%, ResNet 50V2 99.46%, and DenseNet121 99.60%.