Deep Learning-Enabled Hyperspectral Classification of ham

Food authenticity is a crucial aspect of consumer protection, food safety, and quality assurance. Conventional methods for meat authentication often require destructive, time-consuming, or labor-intensive processes. Hyperspectral imaging, which combines imaging and spectroscopy, has emerged as a non-destructive alternative for food classification. This study investigates the application of hyperspectral imaging for differentiating between beef, chicken, and turkey zhambons using one-dimensional convolutional neural networks and long short-term memory networks. Following preprocessing—including segmentation, noise reduction, and spatial averaging—spectral signatures were extracted and classified using deep learning models and then compared to traditional machine learning approaches. The long short-term architecture demonstrated superior performance by effectively modeling sequential spectral dependencies, achieving ۹۹.۹۴% accuracy in the binary classification of chicken versus beef and ۹۸.۱۲% accuracy in the three-class problem (beef, chicken, and turkey zhambons). The findings highlight the potential of hyperspectral imaging combined with machine learning approaches as an efficient tool for processed meat authentication.

کلیدواژه ها:

Convolutional neural network ، Deep Learning ، hyperspectral imaging ، long short-term memory ، machine learning ، processed meat authentication