Analysing cow walking pattern and claw problem data

The researchers have equipped the Dairy Campus innovation centre with cameras and antennas. Cameras in the milking parlour take images of the claws. Cameras at the outdoor area of the milking parlour register the cows’ walking pattern. The researchers then develop an algorithm that detects abnormalities in the walking pattern.

They combine the resulting data with the behavioural data from the living spaces. To combine the video data with behavioural data, the cows get a collar with a sensor. They have also installed cameras in one of the living spaces. Antennas in the living spaces capture the signals from the sensor, enabling them to follow the activities of all cows. The researchers want to analyse and store the data using a modern data warehouse infrastructure that can smart-link video and tracking data.

In the end, farmers with such systems can pick out the cows with locomotion problems earlier and provide the right care. In addition, breeding organisations can use the data for breeding, and select for reduced sensitivity to infectious claw diseases.

Progress (July 2023)

Since December 2022, the infrastructure of the Dairy Campus has been set up to closely monitor a total of 112 animals (in 7 departments) 24/7 using tracking sensors in their living and waiting spaces. These data are used to analyse behaviour in the living space as well gain a better understanding of how infections, such as Digital Dermatitis, are transmitted in a barn. In addition, the installation for the collection of camera footage in the milking parlour was adjusted and improvements were made that allow for making predictions with regard to the presence of Digital Dermatitis as well as linking these predictions to the correct stand number. In order to achieve all this, the researchers worked on a large number of adjustments and expansions to the system allowing for the collection, monitoring and storage of data.

Since September 2022, a number of tools have been developed that can convert the raw data from the cameras and tracking sensors into information or data for further analyses. For example, there is a protocol for the calibration of different cameras, both individually and in stereo. This formed the basis for the development of a tool that allowed for the reconstruction of a 3D point by combining the footage from multiple cameras. The tool can recognise cows in the footage, turn certain parts of the cow into 3D points and track these through time and space. The raw data from the tracking sensors is processed so these positional data can be linked to barn coordinates. The researchers can then link both data streams (position and 3D points) to certain behaviour (ruminating, feeding, drinking).

The processed positional data is also used to develop transmission models for the infectious disease Digital Dermatitis. The presence of Digital Dermatitis is assessed through an automatic detection model that uses cameras installed in the milking parlour. The data provided by the hoof trimmer are the golden standard for the development of this model.

The researchers also worked on the automatic monitoring of the locomotion of animals by gathering data using camera footage from the exit of the milking parlour. These data are analysed to track the walking pattern over time. In future, this can be linked to behaviour in the barn, or the presence or absence of claw problems like Digital Dermatitis.