Evaluation of an Image Analysis Approach to Predicting Primal Cuts and Lean in Light Lamb Carcasses

Abstract

Carcass dissection is a more accurate method for determining the composition of a carcass; however, it is expensive and time-consuming. Techniques like VIA are of great interest once they are objective and able to determine carcass contents accurately. This study aims to evaluate the accuracy of a flexible VIA system to determine the weight and yield of the commercial value of carcass cuts of light lamb. Photos from 55 lamb carcasses are taken and a total of 21 VIA measurements are assessed. The half-carcasses are divided into six primal cuts, grouped according to their commercial value: high-value (HVC), medium-value (MVC), low-value (LVC) and all of the cuts (AllC). K-folds cross-validation stepwise regression analyses are used to estimate the weights of the cuts in the groups and their lean meat yields. The models used to estimate the weight of AllC, HVC, MVC and LVC show similar results and a k-fold coefficient of determination (k-fold-R2) of 0.99 is achieved for the HVC and AllC predictions. The precision of the weight and yield of the three prediction models varies from low to moderate, with k-fold-R2 results between 0.186 and 0.530, p 2 results between 0.080 and 0.461, p < 0.001. The results confirm the ability of the VIA system to estimate the weights of parts and their yields. However, more research is needed on estimating lean meat yield.

Keywords

• light carcass
• cut
• video image analysis
• prediction