Bridging the Gap in Precision Agriculture: A CNN-Random Forest Fusion for Disease Classification

Abstract

Within the framework of a rapidly expanding worldwide population and the critical need to guarantee food security, precision agriculture has arisen as a crucial area of study and advancement. In the scope of this field, our research aims to make a significant impact by enhancing the evaluation of onion smut disease severity through an innovative multiclassification framework. The present study presents a new hybrid model that combines the strengths of Convolutional Neural Networks (CNN) and Random Forest (RF). This model integrates the feature extraction capabilities of deep learning (DL) with the classification robustness of ensemble learning, resulting in a synergistic approach. The combination of many elements leads to the development of a model that not only exceeds current benchmarks but also establishes a notable standard, demonstrating an outstanding overall accuracy rate of 96.38%. The significance of our model extends beyond its exceptional accuracy. The feature interpretability of this confers a significant advantage, as it enables a comprehensive comprehension of the various aspects that contribute to the severity of the condition. The availability of interpretability in this context provides farmers and agricultural specialists with a powerful tool that can significantly enhance their ability to make informed decisions based on data when it comes to managing diseases. Our research represents a groundbreaking advancement in the field of multiclass categorization in the context of agriculture. The historical constraints given by the complexity and diversity of crops and illnesses have been significant. However, our hybrid approach presents a scalable alternative that surpasses the limitations of traditional onion farming. Not only does it offer the potential for improved disease evaluation, but it also establishes a precedent for addressing multiclass classification jobs in the agricultural domain on a wider scale.