Detection and Classification of Cassava Diseases Using Concatenate Model

This study explores the application of a hybrid deep learning approach using an EB0DNet model for cassava leaf disease classification. The CCMT Cassava dataset, which includes a diverse set of disease-affected and healthy leaf images, is utilized to train and validate the proposed model. EfficientNetB0 and DenseNet121 architectures are leveraged for their complementary strengths as EfficientNetB0’s parameter efficiency and DenseNet121’s robust feature extraction capability. By concatenating the feature outputs of these two architectures, the model aims to enhance classification accuracy by combining fine-grained details with broader contextual patterns. The experimental results demonstrate the effectiveness of the concatenate model, showcasing improvements in key performance metrics such as training accuracy with \(99.40\%\) , test accuracy with \(97.61\%\) , precision with \(98\%\) , recall with \(97\%\) , and \(F_1\) -score with \(98\%\) compared to individual network implementations. This research contributes to the field of agricultural disease detection, providing an efficient and accurate tool for early disease identification, which is critical for crop management and yield optimization.