Skip to main content

Computer Vision for Animal Welfare in the Pig Sector

This use case facilitates the assessment of animal welfare in the pig sector by leveraging AgrospAI's data sovereignty capabilities. The assessment is based on video images of pigpens within the farm. Nevertheless, the use of AgrospAI's compute-to-data capability ensures farmers that these images are not leaked and that they maintain control.

Computer Vision services derive welfare metrics from the videos, and data consumers only have access to the metrics after the Computer Vision services are executed on the pigpen images in the compute-to-data environment. Metrics include information about how much each pig has moved (or remained static) or how much time they spend in the feeding or drinking areas of the pen.

Image illustrating individual pig tracking, although the pigpen images are never accessed by consumers, only the calculated welfare metrics.

AgrospAI already has a dataset that includes a sequence of video surveillance images from one of the pigpens at the Centre of Swine Studies of Catalonia (CEP), with CEP being one of the data space participants:

These images can be used by animal welfare assessment algorithms like the one already presented in AgrospAI. It can be run on the pigpen image dataset using compute-to-data provided by Arsys. The algorithm performs automatic image segmentation and tracking to identify and monitor pig movements in the image sequence. It also monitors pig visits to defined areas of interest, such as the automatic feeder or drinker. This allows for the automatic generation of metrics that can be used for animal welfare assessment. The data processing service is available as the