Enhanced Segmentation with Optimized Nine-Layered CNN-Based Classification of Leaf Diseases: An Automatic Approach for Plant Disease Diagnosis

Abstract

The task of monitoring the plant leaves is considered to be error-prone, inconsistent, and unreliable. Thus, certain deep learning algorithms are developed to detect plant leaf diseases, but most deep learning techniques were designed with restricted-resolution images based on convolutional neural networks (CNNs). Hence, this article decides to develop an optimized nine-layer (ONL)-CNN-based plant leaf disease classification (PLDC) model for ensuring accurate results. The leaf segmented images are given into the abnormality segmentation with the “Adaptive Fuzzy C-Means” (A-FCM) technique for getting the abnormality segmented images. The segmented images are inserted into the classification stage, in which the optimized nine layer-CNN is utilized for leaf disease classification. Here, the parameter optimization takes place in FCM and ONL-CNN for enhancing the performance of the developed PLDC model using the developed hybrid heuristic optimization algorithm with hybrid leader cat swarm optimization (HLCSO). The experimental analysis is conducted with the developed PLDC model with different baseline methods to put forward the effectiveness of the developed model. Throughout the analysis, the given designed method achieved a 96% accuracy rate. Therefore, the proposed model ensures its effective performance regarding accuracy metric and also helps to detect early diagnosis of leaf diseases.