Innovative AI model detects damaged pallets at Duvel Moortgat

We always strive to offer our customers a sustainable and innovative solution. Our AI proof-of-concept based on Computer Vision offers a promising solution for Duvel Moortgat.

This concept uses cameras to analyze visual images of pallets and validate whether they still meet quality requirements. Based on the outcome, damaged pallets are automatically removed so that they cannot cause damage in the further process. The AI system not only provides a more efficient way of quality control, but also contributes to the company's forward-looking and innovative approach.

Upon arrival at the warehouse, the pallets at Duvel Moortgat are usually loaded with empties, followed by mechanical inspection. To guarantee the quality of the pallets, they are passed through a visual inspection system. First, they pass a turntable on which two cameras are placed that take images of the front and back of the pallet in a split second. The table then rotates and the two sides are checked for damage. This automated system ensures fast and efficient inspection of the pallets, resulting in higher quality of the production and logistics process. It is possible to integrate this innovative solution into the brewery's existing production line, so there is no disruption to the existing production process.

Next, our AI solution analyzes the images of the pallets and applies a pre-trained classification model to compare the pallets with an ideal image of a pallet in optimal condition. The system validates whether the pallets still meet the requirements and automatically identifies the pallets that should be taken out of circulation.

This approach has several advantages:

During the proof-of-concept, important additional challenges emerged, such as that not all types of defects could be detected with the camera setup used. ACA Group's way of working allowed Duvel to quickly gain insight into the possibilities of computer vision and the conditions for bringing the system into operation for a limited cost.

The proof-of-concept shows that technologies such as Computer Vision have great potential in environments where visual quality control is essential. This applies not only to bottling plants such as Duvel Moortgat, but also to automated warehouses and production processes, among others. By using this technology, processing operations can be carried out faster and more efficiently. The data from image analysis can also be used to recognize patterns and provide input to other areas of the production cycle. For example, materials - in the above case pallets - can be improved to reduce defect rates.