A Custom 1D ResNet-GRU Model for Accurate Pork Adulteration Classification in Beef

Ensuring the consumption of high-quality food and identifying potential adulterants in food products has become essential due to the global prevalence of food adulteration. The detection of adulterants in food, especially pork in beef, poses significant challenges due to ethical, medical, and economic reasons. Conventional techniques often fail to accurately distinguish between pure and adulterated meat, highlighting the need for advanced solutions. This paper introduces a hybrid model combining 1D ResNet and GRU architectures to improve the detection of pork adulteration. By analyzing data from ten e-nose sensors, the model achieves a high classification accuracy of 0.9988 in identifying various levels of pork adulteration, from 0% to 100%. The proposed approach outperforms traditional methods in precision, recall, and F1-score, demonstrating its effectiveness in enhancing food quality assurance, safeguarding consumers, and aligning with regulatory standards.