Brassica Black Rot Severity Levels classification based on Multimodal Convolutional Neural Networks and Support Vector Machines

Brassica black rot is a pressing danger to globally grown brassicas that need more efficient detection and feature severity classification methods. This research proposes a new method by combining Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs) in the classification of Brassica black rot severity into six distinct levels. The proposed multimodal model is trained and evaluated using a carefully selected dataset of 14,200 high-resolution images labeled with severity levels. The CNN component performs especially well in extracting complex features from images and the SVM part does classifications of the extracted features effectively with a 96.73% accuracy rate altogether. The in-depth performance analysis, provided as Table I, explains the model’s accuracy, recall, and F1 score for each severity class that justifies its capability of accurate classification according to different levels. The proposed model, which can be seen in Figure 3 as the latest comparison is regarded to be one of the front-runner solutions when compared with competitive technologies. The classification into six severity levels not only helps to improve accuracy in estimating the disease but also makes it possible for treatment targeted at different stages of the disease. This phenomenon contributes greatly to the sphere of plant pathology as it moves on bettering agriculture technology having a revolutionary solution for accurate assessment of Brassica black rot severity and consequently, improved management practices.