An Image-Based Automated Potato Leaf Disease Detection Model

Production loss in agriculture due to different diseases occurring in plants is a common problem. As traditional identification methods are time-consuming, it is better to have an automated model for the same. Detection of potato leaf diseases has been proposed in this research. Samples of healthy potato leaves, and two diseases early blight, and late blight have been considered. Data augmentation and image segmentation have been used to obtain better accuracy. SURF, Gabor Filter, Wavelet features, and fractal dimension of the images have been extracted as the features. A widely used Plant Village Dataset has been taken to train and test the model. The accuracy of the model is 95.97%. The model outperforms several cutting-edge models in comparison.