Currently pork marbling is assessed subjectively in the industry, because of the limited methods and tools that are suitable for the industry. In this dissertation, we are devoted to develop a computer vision system for objective measurement of pork which suits the industrial needs. Experiment 1 examined the possibility of using computer vision system (CVS) to predict marbling in a lab-based experiment using pork samples that were already trimmed of subcutaneous fat and connective tissue. Experiment 2 an industrial scale CVS was built to predict the 3rd and 10th rib pork chop’s marbling. Experiment 3 the industrial scale CVS was tested in the meat plant and images of whole boneless pork loin were collected. The CVS predicted marbling were compared with subjective marbling score using crude fat percentage (CF%) as standard. In experiment 1 subjective marbling score had a correlation of 0.81 with CF% while CVS had a 0.66 correlation. CVS has shown an accuracy of 63% for stepwise regression model and 75% for support vector machine model. These results indicate that CVS has the potential to be used as an tool to predict pork intramuscular fat (IMF)%. In experiment 2 the accuracy of CVS predicting pork chop CF% was 68.6% and subjective marbling was 70.1%. A drop of accuracy in predicting anterior chop CF% for both CVS and objective marbling score was observed when compared to posterior chop, this suggest that there is a discrepancy in accuracy between the anatomy location of samples collected. In experiment 3 the accuracy of CVS predicting boneless whole loin was 58.6% and subjective marbling score was 53.3%. In this research, CVS has demonstrated a consistency of accuracies using different pork samples. CVS has shown higher accuracy when predicting whole boneless loin IMF% when compared to subjective assessment.