Development of computer vision systems for the inspection of packaging straps

What was the challenge or problem to solve?

Our client needed to develop an artificial vision system for the inspection of packaging straps, aiming to improve the correct application of the plastic straps used to seal their cardboard boxes. The inconsistency in the application of these straps, whether due to breakage or poor adhesion, required an automatic and efficient solution.

The product development team took on the task of developing an advanced computer vision system.

Development of computer vision systems for the inspection of packaging straps

The goal was to obtain a technology capable of detecting the absence of one or both straps with an accuracy greater than 99%. This challenge not only involved the creation of a highly reliable system but also overcoming the difficulties inherent in a production environment, such as variations in lighting and the position of the boxes. Overcoming these obstacles was crucial to ensure the generalization and effectiveness of the system under real working conditions.

How was it addressed or what was the solution?

The team started the project with a detailed study of the case. This included visits to the factory to inspect the production line and meetings with users to define the system’s specifications and requirements precisely. This phase of analysis and planning was fundamental to establish a clear path towards the solution.

Subsequently, the necessary software was developed, using Python and specialized libraries such as TensorFlow and PyTorch.

These tools, known for their effectiveness in artificial intelligence and computer vision projects, were essential for creating software that managed both the camera and the deep learning process.

Finally, the computer vision system was deployed on the production line. This involved the installation of specific hardware, such as industrial cameras and a computer with a GPU capable of performing fast and accurate inference calculations. The implementation of the system in the production environment allowed for final adjustments and ensured its correct functioning.

The ability to detect the lack of straps automatically and accurately has significantly improved the quality of the packaging process, demonstrating the value and innovation that artificial vision brings to the industry.

Contacta con nosotros

Contáctanos para más información o, si lo prefieres, dispones de otras de nuestras consultorías técnicas con servicios complementarios para potenciar tu proyecto.