A novel fine-tuned deep-learning-based multi-class classifier for severity of paddy leaf diseases

Area of Science:

• Agricultural Science
• Computer Science
• Plant Pathology

Background:

• Paddy leaf diseases significantly impact crop yield and quality.
• Accurate detection and severity assessment are crucial for effective disease management.

Purpose of the Study:

• To develop and evaluate a hybrid deep learning model for identifying paddy leaf diseases and their severity.
• To improve the accuracy and efficiency of disease diagnosis in paddy crops.

Main Methods:

• A dataset of 4,068 paddy leaf images was pre-processed and augmented using a generative adversarial network (GAN).
• A hybrid model combining Convolutional Neural Network (CNN) and Support Vector Machine (SVM) was employed for classification.
• Segmentation methods were used for disease severity calculation, with severity levels defined by domain experts.

Main Results:

• The hybrid CNN-SVM model achieved a 98.43% accuracy rate in predicting paddy disease type and intensity.
• The model effectively categorized three major diseases: bacterial blight, blast, and leaf smut.
• The proposed model demonstrated superior performance compared to standalone CNN and SVM models.

Conclusions:

• The developed hybrid deep learning approach is reliable and effective for identifying and assessing the severity of key paddy leaf diseases.
• This method offers a significant advancement over existing classification models for paddy crop monitoring.