Precise Recognition of Adulterated Sliced Mutton Using Machine Vision on Mobile Phone Images.

In recent years, the authenticity of sliced mutton has become a growing concern due to the incorporation of non-mutton ingredients and the increasing use of processed and reconstituted meat products. In this study, a low-cost and non-destructive authentication method integrating smartphone-based image acquisition with machine learning and deep learning techniques was developed for the identification of real, processed, and reconstituted sliced mutton. A total of 600 images were collected under standardized conditions, from which color features in RGB, HSV, and Lab color spaces and texture features derived from the gray-level co-occurrence matrix (GLCM) were extracted. Statistical analyses, including the Kruskal-Wallis test, Dunn's post hoc test, and principal component analysis, demonstrated significant inter-class differences and confirmed the discriminative capability of the extracted features. Four machine learning models (KNN, LDA, RF, and SVM) and three transfer learning-based convolutional neural networks (VGG16, ResNet50, and InceptionV3) were subsequently developed and evaluated. Among the machine learning models, SVM achieved the best classification performance, while VGG16 demonstrated the highest deep learning performance with an accuracy of 96.42 ± 0.51%. Misclassification analysis indicated that processed sliced mutton represented the primary source of classification ambiguity because of its overlapping visual characteristics with both real and reconstituted products. Furthermore, Grad-CAM visualization revealed that the CNN model focused predominantly on texture and structural regions closely associated with muscle and fat distribution, providing interpretability for the classification results. These findings suggest that smartphone-acquired RGB images combined with machine learning and deep learning methods may serve as a promising, low-cost screening approach for preliminary sliced mutton authentication in market surveillance and supply chain inspection.