Goal: This examine evaluated the efficiency of a deep-learning-based mannequin that predicted cooking loss within the semispinalis capitis (SC) muscle of pork butts utilizing hyperspectral photos captured 24 h postmortem.
Strategies and outcomes: Principal part regression (PCR) and partial least squares regression (PLSR) fashions for predicting cooking loss in SC muscle confirmed greater R2 values with multiplicative sign correction, whereas the primary spinoff resulted in a decrease root imply sq. error (RMSE). The deep learning-based mannequin outperformed the PCR and PLSR fashions. The classification accuracy of the fashions for cooking loss grade classification decreased because the variety of grades elevated, with the fashions with three grades reaching the best classification accuracy. The deep studying mannequin exhibited the best classification accuracy (0.82). Cooking loss within the SC muscle was visualized utilizing a deep studying mannequin. The pH and cooking lack of the SC muscle had been considerably correlated with the cooking lack of pork butt slices (−0.54 and 0.69, respectively).
Conclusion: Due to this fact, a deep studying mannequin utilizing hyperspectral photos can predict the cooking loss grade of SC muscle. This implies that nondestructive prediction of the standard properties of pork butts could be achieved utilizing hyperspectral photos obtained from the SC muscle.
Kyung Jo, Seonmin Lee, Seul-Ki-Chan Jeong, Hyeun Bum Kim, Pil Nam Seong, Samooel Jung, Dae-Hyun Lee, Cooking loss estimation of semispinalis capitis muscle of pork butt utilizing a deep neural community on hyperspectral knowledge, Meat Science, Quantity 222, 2025, 109754, ISSN 0309-1740. doi: 10.1016/j.meatsci.2025.109754
