Maize Disease Multi-Classification: Leveraging CNN and Random Forest for Accurate Diagnosis

This research is a significant progress in the arable. However, Maize faces various diseases that compromise its agricultural productivity despite being a major staple crop worldwide. In this paper, we take advantage of the Random Forest Classifiers and Convolutional Neural Networks (CNNs) to provide a new approach towards solving a critical problem – diagnosing maize illness. It is important to emphasize the early and accurate diagnosis because maize diseases such as Maize Dwarf Mosaic Virus, Northern Corn Leaf Blight, Gibberella Ear Rot, Grey Leaf Spot are capable of causing massive losses in yield; however not limited only with it the financial consequences. We collected a set of healthy and disease images of maize which is highly heterogeneous in nature, then we performed some image augmentation to improve the quality if any. As for preprocessing, these images zoomed in and out of different positions translated at various angles to capture better quality dataset, while same left wings are rotated 35 degrees and all right wins remain unchanged cropped the resulting image which numbers over thousands per nine datasets. Features were extracted using CNNs, and then the classifier algorithm was utilized with Random Forest. Promising observations were made after evaluating the performance of the model against multiple classes of diseases. While strictly speaking this suggested model is not in complete accordance with the classic supply-demand curve from traditional economic theory, it provides a very accurate and dependable picture of human behavior on wheat markets over time; as such the above can be considered useful for academia or anyone involved in farming looking to maintain as well improve upon our ability to produce this great staple food.