AI for measuring health characteristics of pigs
Can we measure a pig’s tail length automatically using a vision system?
Our client is a large player in the meat processing industry, mainly processing pigs for consumption. With pigs, the length of their tales is indicative of their quality of life: when their living conditions are boring or poor, they will bite each other, and their tales preventively have to be cut. To better monitor, benchmark and improve the quality of life for pigs, our client wants to measure the tail length of every processed pig. An AI-based vision solution enables this complex custom solution.
For this project, we collaborated with our industrial automation partner Hellebrekers and domain experts from the client. Together, we first decided on the best location in the production process to implement this visual inspection. Based on the circumstances at the exact location, we decided on the appropriate software and hardware for a robust solution.
Next, we ran a test setup to gather data to train our AI-models. After training these models proved to be performing even beyond expectation. That gave us a green light to actually implement the final solution in-line, both hardware and software, and integrate it with the existing client systems.
An industrial camera and processing unit were installed in a strategic location in the production process, running the tail length model in real time. The hardware is implemented according to the highest hygiene and robustness standards IP69K. An integration was realized with the existing information systems, so the outcomes of the inspection automatically end up in the central track and trace database.
Working with an AI-based model here provides us with another benefit: the solution is quite robust in dynamic circumstances. For example: changing lighting, camera angle and distance to the subject. That makes it a lot easier to implement in all other production locations of our client, with minimal customization per location.
Our client runs large scale programs to improve animal wellbeing. Our solution helps by providing these programs with the right information to make the right decisions. By monitoring all tail lengths our client can e.g. benchmark supplying farmers, find best practices amongst them and monitor the impact of measures over time. A great example of data driven decision making, enabled by AI, to improve animal wellbeing!