The live pig appearance identification and prediction system developed in this project has a determination coefficient (R²) as high as 0.78, which is better than the industry average of 0.64. Furthermore, the difference between the predicted carcass and meat quality grades and those obtained by the Soxhlet extraction method is controlled within ±0.2%. This project employs "three-free" technology to construct a complete data chain for objective grading.

The vision of this technology is to achieve "real-time identification of the quality of live pig meat". At the transaction level, it allows farmers and buyers to conduct transactions in a transparent manner by grasping the carcass condition at the farm. At the production level, it allows farmers to dynamically adjust feed formulas to achieve "precision feeding" by providing data feedback on the growth conditions of pigs, thereby increasing lean meat percentage and reducing ineffective feed loss.

This project integrates the feeding, slaughtering and consumer. (1) Lean meat percentage identification system for pig carcasses is installed on the slaughtering line to optimize the identification accuracy and match the slaughtering speed. (2) Pork quality and grading identification system classifies and identifies the lean meat percentage of meat products. (3) Lean meat percentage identification system for live pigs is built at the farm end to adjust the feed formula during the feeding