Westfort and Info Support Realize Reliable AI Application

During Westfort’s processing, the pigs are classified. This includes determining whether the pig is a boar, borg or gelt. Until recently, this classification was done only by humans.

Info Support worked with Westfort to design and implement a solution that uses cameras and Azure Cognitive Services to classify pigs into the three varieties in real time. A snapshot is taken every 5.33 seconds, which is automatically classified via machine learning. This uses edge computing combined with the computing power of the cloud. Because the snapshots are stored with classification, follow-up analysis is possible; an important step in Westfort’s data maturity.

This is a continuous process, says Niels Naglé, data solutions consultant at Info Support. “The environment changes, errors may come up that require you to improve the model, something where human expertise is still very important and remains so.” When training the AI model, first and foremost it is important that you use a good camera, which you place in the right way so that a constant quality image is available to the model. “In addition, you have to take into account the environment. For example, that a carcass is not hanging with the back side facing the camera.

By significantly increasing the overall reliability of the classification, Westfort saves costs, as fewer checks are required and the number of adjustments to records due to errors is minimized. On top of that, the data can now be integrated into Westfort’s data environment, which also makes it possible to check processes after the fact. Ensuring food safety is an important aspect within the food sector and at the same time a major challenge. As a food producer, you need to be able to trace products quickly within the entire chain. In addition, it may be important to provide supporting evidence afterwards. The AI solution that automatically takes snapshots and stores them appropriately makes this possible.

Info Support and Westfort worked together in a mixed team. “We fully included the people at Westfort in what we were doing. Through this co-creation, domains, business knowledge and expertise in artificial intelligence and data came together in an effective way,” Naglé said.