Advanced Multi-Scale Classification of Onion Smut Disease Using a Hybrid CNN-RF Ensemble Model for Precision Agriculture

The production of onions is confronted with a notable challenge in the form of onion smut disease, which is caused by the fungal pathogen Urocystis cepulae. This disease has the potential to cause considerable reductions in crop productivity. This study presents a novel approach for accurately categorizing the levels of onion smut severity scales. The proposed method combines a hybrid Convolutional Neural Network (CNN) and Random Forest (RF) ensemble model. The dataset, which has been carefully annotated to encompass a wide range of disease scales, is subjected to preprocessing and augmentation techniques in order to improve the model’s capacity for generalization. The integrated CNN and RF model has a high level of accuracy, reaching 95.66%, as observed throughout a thorough evaluation process. The suggested technique demonstrates greater performance compared to baseline models, specifically in its ability to accurately capture the nuanced changes in onion smut severity. The utilization of confusion matrices in visualizing data enhances the interpretability of the information, hence facilitating a more thorough comprehension of the model’s capabilities and areas for future improvement. While commemorating these achievements, the study notes the obstacles associated with wider applicability and proposes future avenues for research. This study makes a significant contribution to the integration of computer vision and agricultural science by introducing a realistic and economical solution for accurately classifying onion smut scales in real-world agricultural settings.