Development of a data planning framework for part assessment with Computer Vision in context Circular Economy.

Products returning through higher-level R-Strategies exhibit lower resource consumption and improved sustainability compared to lower-level strategies. To implement these strategies effectively at an industrial scale, an objective and automated assessment of parts is required to determine the highest possible R-Strategy. A promising solution gaining recognition is the use of Computer Vision technology. In traditional linear manufacturing, Computer Vision is already established for automation and quality control. This article investigates which product characteristics can be optically assessed and explores Computer Vision methods that can automate these assessments. Both conventional image processing and ML-based techniques are considered, given the high variability and uncertainty in the condition of returned circular products. The need for extensive data collection, starting early in the product lifecycle, is critical to enable these methods. The findings are incorporated into a framework for practitioners, providing guidance on visual part characteristics, Computer Vision strategies, and data planning for enhanced Circular Economy implementation.