Development of an intelligent meat spoilage detection and grading system using particle swarm optimization-based convolutional neural network

Abstract

This research developed an intelligent meat spoilage and quality grading system using a particle swarm optimization-based convolutional neural network. It addressed the problems associated with the subjective manual assessment of meat quality and inefficient and expensive meat quality grading systems as well as the lack of a comprehensive dataset for meat quality detection. This research created a new dataset for meat spoilage and quality detection. Furthermore, a PSO-based convolutional neural network was trained with the new dataset for the classification and the grading of the meat. The Python code is then integrated into the Raspberry Pi 4 to make it a stand-alone system. Comparative analysis indicated that the PSO-based CNN performed better compared to the baseline CNN by 2.91% for accuracy, 2.49% for precision, 0.99% for F1-score, 1.87% for recall, 2.74% for specificity and 1.14% for sensitivity. The obtained results implied improved food safety in the food processing industry and retail environments. In addition, the intelligent system provides support to human experts for accurate assessment of meat quality.