The assessment of cocoa bean quality plays a crucial role in the chocolate industry, and automated approaches utilizing image processing techniques and classifiers have become increasingly appealing. In this study, we implemented and compared the performance of image classifiers using Convolutional Neural Network (CNN) architectures for cocoa bean quality classification. By employing this approach, we developed a system capable of accurately and efficiently classifying cocoa bean images, reducing dependence on human evaluation. We compared several CNN architectures, including VGGNet, to evaluate their performance in cocoa bean image classification. Experimental results demonstrated that CNN-based classifiers can provide accurate assessments of cocoa bean quality, with significant success rates. This research contributes to the development of efficient and accurate image classification systems for cocoa beans, which can enhance efficiency in the chocolate industry and ensure product quality. Additionally, our testing results indicate that the model with a batch size of 64 achieved the highest accuracy of 98.44%, outperforming the other three tested batch sizes in cocoa bean classification performance.
Copyrights © 2023