Development and evaluation of Video Image Analysis (VIA)

Sustainable Livestock Production LINK Programme
Project LK0673F: Development and evaluation of Video Image
Analysis (VIA) systems to monitor growth and carcass quality in
live beef animals
Start Date: 01/04/2007
End Date: 30/09/2008
Executive summary report
A feasibility LINK project was undertaken to evaluate video image analysis
systems for the assessment of beef cattle (BEEFVIA) during an on-farm
finishing period and within a commercial auction market situation. Two
experiments were carried out during which dimensional datasets were
gathered from live beef cattle using digital video cameras and assessed as
potential predictors of commercially important finishing and carcass
characteristics.
The first experiment involved gathering BEEFVIA images from 48 steers of
defined genotypes during a 2-3 month finishing period when the steers were
housed and fed under commercial farming conditions. BEEFVIA equipment
designed as part of this project proved practical, reliable and robust within a
typical farming environment. Prediction equations derived from BEEFVIA
dimensional data produced acceptably low levels of predictive error (+/- 2.6 to
8.0 % of actual values) for slaughter characteristics such as liveweight (LW)
and cold carcass weight (CCW) at slaughter, sirloin weight and sirloin
proportion at slaughter and killing out proportion (KO) in the 48 steers under
study. The feasibility of BEEFVIA dimensions to predict parameters such as
fat and conformation grading at slaughter (particularly using neural network
analysis) and liveweight gain throughout the finishing phase was also
demonstrated in this study. However, more extensive studies are required to
improve the predictive accuracy of the system for these important
characteristics.
The second experiment involved gathering BEEFVIA images from 49 steers
and heifers of mixed genotypes that were marketed through a commercial
auction market on 12th May 2008. Again the BEEFVIA equipment proved to
be both reliable and robust with dimensional datasets being captured for all 49
animals.
No impediments to the smooth running of the sale were
encountered. Examination of the prediction equations developed during the
first “on-farm” experiment suggested that situation specific prediction
equations may need to be developed for either the “on-farm” or “auction
market” scenarios separately as a result of likely differences in the relative
dimensional predictors between animals in each of these respective
environments. However, prediction equations with acceptably low levels of
predictive error (+/- 4.2 to 8.0 % of actual values) for slaughter characteristics
such as LW, CCW and KO at slaughter could also be derived from BEEFVIA
dimensional data.
It is concluded the live animal BEEFVIA management systems are practical
and can feasibly be used in the monitoring and assessment of beef cattle
during firstly an on-farm finishing period and secondly for the prediction of
slaughter and carcass characteristics within a working auction market.
BEEFVIA management systems should now be developed as a management
tool for beef cattle farmers during the finishing phase of the production cycle
so that improvements in business efficiency and reductions in waste and
gaseous pollutants can be achieved. Distinct auction market BEEFVIA
systems should also be developed to improve the point of sale information
available to both buyers and sellers of finished cattle within commercial
auction markets. This would facilitate improved market signals throughout the
beef supply chain and provide mechanisms to reward efficient producers and
suppliers of quality cattle that met market requirements more closely.
CONTACT
Dr Jimmy Hyslop
SAC Select Services
FBS Area Office
Bush Estate
Midlothian
EH26 0PH
Jimmy. [email protected]
PARTNERS
QMS
Scotbeef Ltd
Osbourne (Europe) Ltd
John Swan Ltd
Silsoe Livestock Systems Ltd
GOVERNMENT SPONSORS
Defra