Running summarizing tools on a new article

Abstract: In this paper, we presented an approach to deploying a student–teacher feature pyramid model (STFPM) for anomaly detection metal-textile gas filters used in automotive exhaust gas filtering at GKD-Gebr. Kufferath AG. As the customer requires 100% quality of the delivered parts, an optical inspection process of every produced filter is required. This is very demanding for the human inspection worker as she has to inspect many 100 parts in an 8 hours shift. On the other hand, a fully vision-based system is not able to achieve the required classification rates either. Therefore, we propose a one-class anomaly detection process for the gas filters where human and AI work together in achieving the 100% pass rate. The STFPM model deals with the large amount of clearly true positive cases and automatedly classified them as PASS. Only cases of doubt where an anomaly has been detected are inspected by the human inspector. This way, the work load of the inspection worker is reduced, and, on the other hand, the hard to meet case of no mis-classification of the AI system can be avoided. We show the network architecture and the integration into the quality inspection process of the company GKD.