An Efficient Deep Learning Model for Detection of Rice Leaf Diseases using Paddy Doctor Dataset

Detecting diseases in rice leaves is vital for ensuring agricultural productivity and food security. This paper introduces a highly efficient deep learning model designed for the automated identification of rice leaf diseases. Traditional diagnosis methods require expert supervision, which is laborious and costly. We employed an enhanced deep learning model which is a modification over ResNet50 model, classify images of paddy leaves into nine disease categories (bacterial leaf blight, leaf streak, panicle blight, blast, brown spot, dead heart, downy mildew, hispa, tungro), plus a category for healthy leaves. Using the Paddy Doctor dataset with around 24,000 images augmented from the original 16,225, we trained and fine-tuned models to achieve high accuracy despite the complexity of ten classification categories. The proposed deep learning model was adapted with the goal of attaining precise and accurate identification of paddy diseases, the highest achieved accuracy being an impressive ${9 8. 0 3 \%}$.