Stewart S M, Gardner G E, Williams A, Pethick D W, McGilchrist P, Kuchida K
Advanced Livestock Measurement Technologies (ALMTech), Murdoch University, School of Science, Health & Engineering, Western Australia 6150, Australia.
Advanced Livestock Measurement Technologies (ALMTech), Murdoch University, School of Science, Health & Engineering, Western Australia 6150, Australia.
Meat Sci. 2021 Nov;181:108369. doi: 10.1016/j.meatsci.2020.108369. Epub 2020 Nov 11.
This study assessed the precision and accuracy in the prediction of chemical intramuscular fat (IMF%), Meat Standards Australia (MSA) marbling score and AUS-MEAT eye-muscle area (EMA) using Meat Imaging Japan (MIJ) prototype camera systems. Eleven carcass datasets from the Beef Information Nucleus (BIN) project were compiled with carcass grading, IMF% and camera data. Camera prediction of IMF%, MSA marbling score and EMA was assessed using a leave-one-out cross validation method. There was an association between MIJ mirror and MIJ-30 camera traits and IMF%, MSA marbling score and EMA. However, for both prototypes precision varied for IMF% (R = 0.4-0.5, RMSECV = 1.5-1.6%), MSA marbling (R = 0.3-0.5, RMSECV = 57.5-59.3) and EMA (R = 0.7-0.6, RMSECV = 4.1-5.8 cm). Accuracy also fluctuated with average bias values of 1.7-1.8%, 45.8-40.0 units and 3.8-4.1 cm for IMF%, MSA marbling score and EMA respectively. Key differences between carcass and camera traits and processing factors affecting the grading site are likely to have contributed to this variation.
本研究使用日本肉类成像(MIJ)原型相机系统评估了化学肌内脂肪(IMF%)、澳大利亚肉类标准(MSA)大理石花纹评分和澳大利亚肉类眼肌面积(EMA)预测的精度和准确性。收集了来自牛肉信息核心(BIN)项目的11个胴体数据集,包括胴体分级、IMF%和相机数据。使用留一法交叉验证方法评估了相机对IMF%、MSA大理石花纹评分和EMA的预测。MIJ镜面和MIJ - 30相机特征与IMF%、MSA大理石花纹评分和EMA之间存在关联。然而,对于这两种原型,IMF%的精度有所不同(R = 0.4 - 0.5,RMSECV = 1.5 - 1.6%),MSA大理石花纹(R = 0.3 - 0.5,RMSECV = 57.5 - 59.3)和EMA(R = 0.7 - 0.6,RMSECV = 4.1 - 5.8 cm)。精度也有所波动,IMF%、MSA大理石花纹评分和EMA的平均偏差值分别为1.7 - 1.8%、45.8 - 40.0单位和3.8 - 4.1 cm。胴体和相机特征之间的关键差异以及影响分级部位的加工因素可能导致了这种变化。