Comparison of innovative and non-invasive methods in estimating the fat content in pork trimmings

Abstract

The purpose of research was to determine a possibility of application of computer vision systems (CVS) for estimation of fat content in pork trimmings in comparison with methods based on DXR (dual energy X-ray) and NIR (near-infrared reflectance spectroscopy). Research was conducted on 232 samples of pork trimmings. In order to verify the fat content determined by CVS, DXR, and NIR methods, fat content was also determined by the Soxhlet reference method. It was found that CVS can be used to estimate fat content in pork trimmings with a standard error of prediction between 4.9 and 5.6%. In order to achieve higher efficiency, it seems advisable to grind and standardise meat in a meat grinder with a kidney shaped plate.