Integrating Convolutional Neural Networks and Random Forest for Accurate Grading of Rice Spot Disease Severity

Rice Spot disease is known to cause a severe problem that not only affects yield quality and quantity but also poses a considerable challenge to rice farmers all over the world. The traditional method which uses eyesight visual attribution intuitively all time-consuming, subjective, and often inaccurate. This study presents the initial implementation of a novel hybrid model integrating convolutional neural network (CNN) and Random Forest algorithm that may serve to effectively evaluate the spot disease severity grade on the target with high accuracy. The proposed methodology involves gathering exhaustive data, meticulous preprocessing, and the integration of a CNN into a feature-extracting process as well as classification, that applies Random Forest, as a solution to the issue brought about by the existing diagnostic approaches. Applying the method to a varied dataset verges a 95.83% precision rate, clearly beat as well as landline model. Moreover, this model competes at the rank of state-of-the-art. This effort in the first place provides an effective tool for precision agriculture and then could serve as a precedent to contribute to the wider acceptance of artificial intelligence in crop disease management. The achievement of the hybrid system evidences the fact that ML technologies can cause a ‘revolution’ in the agriculture field through the shift to scale and effective solutions to the critical challenges of food security.