Accurate crop disease identification and classification in smart agriculture using a three-tier model and optimized fully conventional network

Identification and classification of crop diseases are essential for efficient agricultural management because they allow early procedures to reduce crop losses and boost yields. Reduced food production, significant crop damage, and financial losses might result from inaccurate or delayed disease identification. To address these challenges, this work focuses on developing a hybrid deep-learning approach for the detection and classification of crop diseases. By integrating advanced techniques from deep learning, this model aims to enhance the accuracy and efficiency of disease identification in smart agriculture systems. The proposed model comprises four main phases: data collection, pre-processing, segmentation, feature extraction, feature selection, and disease detection and classification. Gaussian-adaptive bilateral filters are utilised in the pre-processing stage to reduce noise and blur, and histogram equalisation techniques are employed to improve the quality of the images. The segmentation step utilizes the Optimized Fully Convolutional Network (FCN) to isolate plant leaves from the background, while fine-tuning the model using the hybrid optimization approach Eclectic Harmony Optimization (EHO) ensures accurate differentiation of healthy and diseased areas. Feature extraction involves employing texture analysis techniques and shape descriptors to capture distinct characteristics of crop diseases. The proposed model is implemented using PYTHON. The proposed model outperformed advanced methods with an accuracy of 98.93%, demonstrating its competitive performance in identifying agricultural diseases.