Abstract
Application: The Beef Monitor system accurately monitored individual animal growth and predicted liveweight, cold carcass weight, saleable meat yield and EUROP classifications through passive weighing and 3D imaging of live cattle.
Introduction: Beef producers currently assess the performance of their cattle through visual assessment or by weighing through a crush. This can lead to animals being retained on farm too long, thus not achieving the optimal market price and increasing production cost to the farmer. Video imaging analysis (VIA) is increasingly used in abattoirs to grade carcases and there is potential for 3d imaging to be used on farm to predict carcass characteristics of live animals. The Beef Monitor (BM) system combines a water trough with a non-intrusive automated weighing system and 3D camera technology to track growth and predict carcass characteristics of live animals. The objectives of this study were to validate the use of the BM system and to develop artificial neural networks (ANNs) to predict liveweight and carcass characteristics from images of
live animals. Material and methods: To validate the accuracy of the weigh crate 39 steers were weighed weekly in a crush and these weights were related to the average of all weights recorded on the same day for each beast by the BM crate. Seven BM systems were installed on commercial and research farms in Scotland and a variety of breeds (steers and heifers) were placed behind the system for 1-3 months pre-slaughter. Images and weights were passively collected from individual
animals at each visit to the trough. A further trial was conducted at the abattoir where live animals were weighed and 3D images taken immediately pre-slaughter. Cold carcass weights (CCW) were provided by the abattoir, and saleable meat yield (SMY), fat and conformation grades were determined by VIA of carcass images. ANNs were developed to predict liveweight and carcass characteristics from 40 measurements obtained from the 3D image (heights, widths and lengths) and 20 calculated areas, volumes and ratios. ANN performance was assessed by regression for liveweight, CCW and SMY, and prediction of the correct EUROP grade for fat and conformation.
Results The relationship between weights of 39 steers measured weekly in the crush and the average of weights measured in the BM crate on the same days had an R2 of 0.99 (n = 224, Figure 1). Predictions for all animals, including sex and breed as factors gave the following R2 values: Liveweight R2 = 0.72 (n = 40930), CCW R2 = 0.91 (n = 1655) and SMY: R2 = 0.80 (n = 1655, Figure 2). Prediction of EUROP fat grade had 63% accuracy, and for EUROP conformation grade prediction accuracy was 69% (n = 1655).
Conclusion: The BM system can provide accurate weights for individual animals on a daily basis without the need for manual handling. 3D imaging of live cattle can be used to predict carcass characteristics on farm, presenting an opportunity to improve the efficiency of beef production enterprises through marketing of animals at the optimal time.
Introduction: Beef producers currently assess the performance of their cattle through visual assessment or by weighing through a crush. This can lead to animals being retained on farm too long, thus not achieving the optimal market price and increasing production cost to the farmer. Video imaging analysis (VIA) is increasingly used in abattoirs to grade carcases and there is potential for 3d imaging to be used on farm to predict carcass characteristics of live animals. The Beef Monitor (BM) system combines a water trough with a non-intrusive automated weighing system and 3D camera technology to track growth and predict carcass characteristics of live animals. The objectives of this study were to validate the use of the BM system and to develop artificial neural networks (ANNs) to predict liveweight and carcass characteristics from images of
live animals. Material and methods: To validate the accuracy of the weigh crate 39 steers were weighed weekly in a crush and these weights were related to the average of all weights recorded on the same day for each beast by the BM crate. Seven BM systems were installed on commercial and research farms in Scotland and a variety of breeds (steers and heifers) were placed behind the system for 1-3 months pre-slaughter. Images and weights were passively collected from individual
animals at each visit to the trough. A further trial was conducted at the abattoir where live animals were weighed and 3D images taken immediately pre-slaughter. Cold carcass weights (CCW) were provided by the abattoir, and saleable meat yield (SMY), fat and conformation grades were determined by VIA of carcass images. ANNs were developed to predict liveweight and carcass characteristics from 40 measurements obtained from the 3D image (heights, widths and lengths) and 20 calculated areas, volumes and ratios. ANN performance was assessed by regression for liveweight, CCW and SMY, and prediction of the correct EUROP grade for fat and conformation.
Results The relationship between weights of 39 steers measured weekly in the crush and the average of weights measured in the BM crate on the same days had an R2 of 0.99 (n = 224, Figure 1). Predictions for all animals, including sex and breed as factors gave the following R2 values: Liveweight R2 = 0.72 (n = 40930), CCW R2 = 0.91 (n = 1655) and SMY: R2 = 0.80 (n = 1655, Figure 2). Prediction of EUROP fat grade had 63% accuracy, and for EUROP conformation grade prediction accuracy was 69% (n = 1655).
Conclusion: The BM system can provide accurate weights for individual animals on a daily basis without the need for manual handling. 3D imaging of live cattle can be used to predict carcass characteristics on farm, presenting an opportunity to improve the efficiency of beef production enterprises through marketing of animals at the optimal time.
Original language | English |
---|---|
Publication status | Print publication - 2018 |
Event | BSAS Annual Conference - Dublin, Ireland Duration: 9 Apr 2018 → 11 Apr 2018 |
Conference
Conference | BSAS Annual Conference |
---|---|
Country/Territory | Ireland |
City | Dublin |
Period | 9/04/18 → 11/04/18 |