ABSTRACT
The proposed work detects plant leaf diseases in paddy, where preprocessing involves Laplacian filter and adaptive histogram equalization (AHE) to improve the quality of images. Fusion background removal is used along with HSV color mode and thresholding. A suitable data augmentation techniques using ANN-GAN is used to attain a balanced dataset. Feature extraction involving geometric transformations, contrast adjustments, brightness modifications, and saturation modifications are used to improve the feature extraction ability. Disease spots are segmented using KNN on augmented images and finally the classification stage employs a DBN-ACO approach, combining Deep Belief Networks (DBN) for feature extraction and Ant Colony Optimization (ACO) for classification. These results show an accurate disease classification than other existing methods.
