In the food processing industry, traceability of products is essential for ensuring safety and quality, as well as for enabling a rapid response in the event of recalls. To achieve traceability and detect process deviations in the sterilization area of a production line, we propose a computer vision system. It provides reliable pallet detection, tracking within an individual camera, and the transfer of identifiers between multiple cameras. Within the scope of this thesis, we evaluated existing methods of object detection and tracking and added logic for managing identifiers throughout the entire production process. The system was validated in four stages: first on a simplified model, then in a simulation environment, subsequently in a real multi-camera test environment, and finally in the production facility. Its performance was evaluated through automated tests of various scenarios and standard tracking metrics.
|
