KI-Photoline
KI-Photoline
Photogrammetric 2D/3D inspection system using neural networks for in-line anomaly detection
Photogrammetric 2D/3D inspection system using neural networks for in-line anomaly detection
Project Summary
Project Summary
The joint project “KI-PhotoLine” is researching a novel approach to rapid optical inline quality control. This involves the use of intelligent global 2D defect detection in combination with local one-shot 3D defect measurement.
The joint project “KI-PhotoLine” is researching a novel approach to rapid optical inline quality control. This involves the use of intelligent global 2D defect detection in combination with local one-shot 3D defect measurement.
Project Partners
Project Partners
What are we doing?
What are we doing?
A production line involves many objects that can have different shapes, materials, and surfaces. The solution to be researched should be able to inspect all objects and then make a sorting decision based on specified quality standards.
Particular attention is paid to the inspection of highly reflective, very dark, and/or matte components. The automated quality control of such industrial products (e.g., painted car body parts) in the process cycle and the requirement for 100 percent inline inspection pose major challenges for existing solutions.
Fraunhofer IPK is focusing its research on the areas of defect detection and pose estimation using machine learning, defect localization in multi-camera systems, research into assistance systems for process optimization, and feeding the expert knowledge gained in the process back into the learning systems.
A production line involves many objects that can have different shapes, materials, and surfaces. The solution to be researched should be able to inspect all objects and then make a sorting decision based on specified quality standards.
Particular attention is paid to the inspection of highly reflective, very dark, and/or matte components. The automated quality control of such industrial products (e.g., painted car body parts) in the process cycle and the requirement for 100 percent inline inspection pose major challenges for existing solutions.
Fraunhofer IPK is focusing its research on the areas of defect detection and pose estimation using machine learning, defect localization in multi-camera systems, research into assistance systems for process optimization, and feeding the expert knowledge gained in the process back into the learning systems.
The KI PhotoLine project “Photogrammetric 2D/3D inspection system using neural networks for in-line anomaly detection” is funded by Investitionsbank Berlin and co-financed by the European Regional Development Fund (ERDF).
The KI PhotoLine project “Photogrammetric 2D/3D inspection system using neural networks for in-line anomaly detection” is funded by Investitionsbank Berlin and co-financed by the European Regional Development Fund (ERDF).
