<p>Image classification involves analyzing various characteristics of an image, such as color, size, shape, and texture. These features can be used with machine learning and computational algorithms to automate the process and improve accuracy. This paper introduced a dynamic mutation factor in the differential evolution algorithm for clustering before classification of vegetable images. The proposed method used bag-of-features (BoF) for image representation in classification tasks. The BoF method converts the image data into visual features, which are then processed using the proposed variant of differential evolution. This approach helps extract essential information from the image dataset, making the classification process more robust and precise. To evaluate the model’s effectiveness, the proposed system was trained, validated, and tested using data sets containing six categories of vegetable images. The performance of the proposed model was compared with existing methods to assess its efficiency and reliability. The classification results proved that the proposed modification improved accuracy, demonstrating that the modified DE algorithm outperformed traditional approaches in vegetable classification tasks.</p>