Prediction of pork quality using online computer vision system

With the increasing demand of high quality of pork products from the global market the United States pork industry needs to seek new solutions for determining quality of pork products. In the current pork supply a great deal of variation exists when considering the eating quality of pork such as lean color, intramuscular fat, and tenderness. The technology validation we are proposing is a step forward in creating an accurate, rapid online tool that pork processors could use to sort, ‘grade’, or ‘certify’ product into categories or quality specifications for customers. This research invented an industry suitable online pork quality assessment system based on computer vision technology. The project results found by using our proposed computer system, pork loin color evaluation (coefficient of determination, R2=0.90) was better than subjective evaluation (R2=0.68), For pork marbling assessment, our computer vision system reached same evaluation level (R2=0.62) as subjective assessment (R2=0.63) method.