DEEP LEARNING BASED RICE LEAF DISEASE CLASSIFICATION USING RESNET

Rice leaf diseases is vital to crop and agricultural sustainability. Within this paper, an application of ResNet-based
classification model towards the classification of three most prevalent rice leaf diseases: Brown Spot, Bacterial
Leaf Blight, and Leaf Smut is introduced. The process begins from preprocessing rice leaf images, varying from
image enlargement, resizing, and normalization in an effort to improve data quality as well as model compatibility.
A pre-trained fine-tuned ResNet model is used to get deep hierarchical features as well as image classification
into disease classes. The data is collected from the Plant Village repository and includes more than 87,000 labeled
images and was divided into an 80:20 ratio for training and validation. Experimental results indicate that the
provided model achieves good performance with 98.45% classification accuracy, precision of 97.77%, recall of
98.00%, and 97.22% F1-score. The confusion matrix indicates that there is less misclassification, proof of the
discriminatory power of the model. Results confirm the stability and robustness of the ResNet model for rice
disease diagnosis. The model has significant potential to be applied in field environments in intelligent farming
systems to offer automatic, fast, and accurate identification of plant diseases, enabling farmers to make timely
decisions for the guarantee of protection of yield and quality