World Congress on Computers in Agriculture 2009 edition:7 location:Reno, Nevada, United States of America date:22-24 June 2009
Lameness, an increasing animal welfare problem, has a negative impact on milk production, body condition and reproductive performance in dairy cows. This paper describes a synchronized system which is useful for automatic lameness detection in dairy cattle.
The system consists of a pressure sensitive mat, two cameras and a separation fence. The pressure mat measures the pressure distribution over time for each hoof of the cow. The cameras record the posture and movements of the cow. The separation fence ensures that there is only one cow passing the corridor with the synchronized system.
When a cow passes the separation fence the recording of the pressure mat is started. The mat sends a synchronization signal to the camera system, which automatically triggers the cameras to start recording. When there is no activity on the pressure mat for some time, a signal is send again to the camera system which stops the recording of the cameras. The recordings of the pressure mat and the cameras are saved together with (relative) timestamps so they can be combined afterwards.
The setup of this system provides us with synchronized data on the pressure distribution under the hoof of the cow in combination with images of the posture of the cow at the same moment in time. When combined in MatLab the information can be used for automatic lameness detection.
Experiments with this system were performed on ILVO farm in Ghent Belgium in October 2008. The system recorded postures and pressure distributions of 66 lactating Holstein cows while passing the corridor after milking. The pressure mat was recording at a rate of 60 frames per second, where the cameras were working at 20 frames per second on average. The average synchronization error between the pressure mat and the camera was less than 16 ms tested on 30 recordings. It is assumed that synchronized data of pressure distribution and posture of cows increases the potential for early online and automatic lameness detection.