Transfer Learning for Rice Plant Leaves Classification

This study suggests a transfer learning-based methodology for categorizing rice leaf diseases. The approach makes use of a pre-trained model that was developed using a sizable picture dataset. The model can take characteristics out of pictures, which are then used to categories the pictures. A collection of pictures of rice leaves was used to train the model. 100 epochs are used to train the model. The predictions for the rice leaf disease are quite encouraging. The total accuracy is fairly good at 0.94. At 0.95, the macro average F1 score is likewise exceptionally high. This indicates that the model can accurately forecast the occurrence of all three illnesses. With flawless accuracy, recall, and F1-score, the model excels in predicting leaf smut. This is hardly shocking considering how distinct and simple it is to recognize leaf smut.