Leaf Disease Detection in Banana Plant using Gabor Extraction and Region-Based Convolution Neural Network (RCNN)

Disease identification in bananas has proven to be more difficult in the field due to the fact that it is susceptible to a variety of diseases and causes significant losses to farmers. As a result, this research provides improved image processing algorithms for earlier disease identification in banana leaves. The images are preprocessed using a histogram pixel localization technique with a median filter and the segmentation is done through a region-based edge normalization. Here a novel integrated system is formulated for feature extraction using Gabor-based binary patterns with convolution recurrent neural network. Finally, a region-based convolution neural network is used to identify the disease area by extracting and classifying features in order to increase disease diagnostic accuracy. The proposed Convolutional Recurrent Neural Network–Region-Based Convolutional Neural Network (CRNN–RCNN) classifier provides a precision score of 97.7%, a recall score of 97.7%, and a sensitivity score of 98.69% when evaluated in a dataset with complex image backgrounds. For the banana dataset, the proposed CRNN–RCNN model achieves an accuracy of 98%, which is greater than the accuracy obtained by CNN (87.6%), DCNN (88.9%), KNN (79.56%), and SVM (92.63%).