"Solution based on Artificial Intelligence and computer vision for inspection mask generation"
Computer vision and AI to accelerate the commissioning of automated inspection systems
TECNALIA and AUTIS, a company specialised in industrial monitoring and automation systems, have worked together on the development of a solution based on Artificial Intelligence and computer vision to automate one of the most complex tasks in the implementation of such systems: inspection mask generation.
Automated paint defect inspection is a key stage in vehicle manufacturing. To ensure the quality of the vehicle bodywork, plants use computer vision systems that are capable of identifying surface defects at various stages of the production process.
Automating the configuration of new systems
Nowadays, the masks defining the inspection zones must be set manually for each vehicle model and for each installation. This process requires specialised staff and accounts for a significant proportion of the work involved in the commissioning of new inspection systems.
- The project has researched the use of AI-based segmentation models capable of generating these masks automatically from images captured by inspection systems.
- As a result, it is possible to significantly reduce the amount of manual tasks required during initial set-up and facilitate adaptation to different models, plants and production lines, in addition to a more precise definition of the inspection masks, which will lead to more accurate inspections.
A solution validated in real industrial environments
To develop the technology, a set of more than 650 images from six manufacturers, twenty vehicle models, eight plants and thirteen different production lines was used. This diversity has enabled the solution’s ability to adapt to different industrial environments to be assessed.
- The results obtained demonstrate that automatic mask generation via computer vision and Artificial Intelligence is technically feasible and reduces the effort required to set up new inspection systems.
- Furthermore, the research carried out has highlighted the importance of having standardised labelling criteria in order to continue improving the generalisation of the models.
Towards more agile and scalable roll-outs
The progress made will accelerate the roll-out of new automated inspection facilities, reduce reliance on specialised manual tasks and facilitate the scalability of AUTIS solutions across different plants and vehicle models.
This collaboration reinforces the potential of Artificial Intelligence as a tool to optimise the industrialisation of computer vision systems and move towards increasingly efficient and flexible manufacturing processes.