Prediction of Pork Belly Composition Using the Computer Vision Method on Transverse Cross-Sections

The objective of this study was to identify the pig belly characteristics and to develop regression equations predicting its composition. Based on video image and chemical analysis of 216 bellies, the predictive variables were selected according to their relation to chemically determined belly lipid contents. To estimate the belly fat percentage (BF%), the two best equations constructed were: Equation 1: BF% = 49.960 - 0.7174 × SHME2 + 0.5047 × HE2A (R² = 0.66, RMSE = 3.22); Equation 2: BF% = 43.888 - 0.6014 × SHME2 + 0.4769 × HE2A + 0.0014 × ARTO2 - 0.2697 × HE3A (R² = 0.70, RMSE = 2.25), where: SHME2 = lean meat percentage area of the belly 2 from total cut area, HE2A = the Belly2 height at point 1, ARTO2 = the Belly2 total cut area, HE3A = the Belly3 height at point 1. Compared to lean meat, the percentage of belly fat (BF%) appears to be a more appropriate criterion for the objective evaluation of belly composition due to the simplicity and accuracy of the final regression equation (higher R²).