2021/10/18 by Patrick Hemmer, Niklas Kühl, Hemmer, Patrick +3
Computer Science · #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #Data Visualization and Analytics #FOS: Computer and information sciences #Machine Learning (cs.LG)
paper · pdf · doi:10.48550/arxiv.2110.09023
openalex publication_date 2021/10/18 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Computer-generated imagery of car models has become an indispensable part of\ncar manufacturers' advertisement concepts. They are for instance used in car\nconfigurators to offer customers the possibility to configure their car online\naccording to their personal preferences. However, human-led quality assurance\nfaces the challenge to keep up with high-volume visual inspections due to the\ncar models' increasing complexity. Even though the application of machine\nlearning to many visual inspection tasks has demonstrated great success, its\nneed for large labeled data sets remains a central barrier to using such\nsystems in practice. In this paper, we propose an active machine learning-based\nquality assurance system that requires significantly fewer labeled instances to\nidentify defective virtual car renderings without compromising performance. By\nemploying our system at a German automotive manufacturer, start-up difficulties\ncan be overcome, the inspection process efficiency can be increased, and thus\neconomic advantages can be realized.\n