Deep Learning Techniques for the Automatic Detection and Classification of Rice Diseases

Several studies have processed leaf image data and protected plants from diseases using machine learning classifiers. Most of the suggested classifiers use manually created features from the images to train and test them on data sets to categorize the leaves. This dataset contains 3355 photos of both healthy and disease-ridden rice leaves of 3 distinct varieties. To train our classifier, we've introduced the CNN, InceptionV3, and EfficientNet learning algorithms. These techniques are transfer learning, which is utilized for deep learning. Each subsequent step in transfer learning is better than the last. To evaluate the suggested deep model, we have used visualization tools to understand symptoms and determine the locations of diseased leaf sections. The performances were better than expected as the CNN, InceptionV3, and EfficientNet models, respectively, achieved accuracy rates of 62.94%, 54.73%, and 72.52%. Farmers can use these results as a useful tool to protect rice from contamination.