Hybridization of ResNet with YOLO classifier for automated paddy leaf disease recognition: An optimized model

Abstract Paddy is the most significant crop utilized by more than 2.6 billion people. The paddy crops are affected by various diseases that are unidentified and reduced the production of crop yield. Nowadays, the plants diseases and pests spread increasingly due to the climate change, trade, and globalization. The plant pathogens can be viral, fungal, nematodes or bacterial that affects all parts of the plants. The challenging tasks are to determine the symptoms and identify the controlling measures of the plant diseases. The plant leaves can be affected by numerous diseases, which results in destruction in terms of crop field to various social and economic aspects. The deep structured architectures and machine learning are implemented in the conventional models for detecting the leaf diseases. Hence, the main intention of this study is to develop the novel model for paddy leaf disease recognition using the hybrid deep learning. Initially, the input paddy leaf images are collected from standard sources that undergo filtering and contrast enhancement approaches. Further, the segmentation of the abnormal region of the paddy leaf is done by “adaptive K‐means clustering.” This is also accomplished by the Fitness Sorted‐Shark Smell Optimization (FS‐SSO). With the segmented images, the recognition of the disease is performed by the hybrid deep learning using the Resnet and YOLO classifier. As the modification, the fully connected layer of the ResNet model is replaced by the YOLO classifier for disease recognition. The significant parameters of the hybrid deep learning are optimized by the FS‐SSO for attaining the high recognition rate. Experimental analysis is performed for computing the performance metrics and the accuracy of the classification for evaluating the efficiency of the suggested method.