2021/10/15 by Jože M. Rožanec, Rožanec, Jože M., Elena Trajkova +7
Computer Science · Decision Sciences · Engineering · #Advanced Statistical Process Monitoring #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Industrial Vision Systems and Defect Detection #Machine Learning and Algorithms #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2110.09396
openalex publication_date 2021/10/15 · openalex created_date 2022/11/14 · openalex updated_date 2026/07/28
Quality control is a key activity performed by manufacturing companies to\nverify product conformance to the requirements and specifications. Standardized\nquality control ensures that all the products are evaluated under the same\ncriteria. The decreased cost of sensors and connectivity enabled an increasing\ndigitalization of manufacturing and provided greater data availability. Such\ndata availability has spurred the development of artificial intelligence\nmodels, which allow higher degrees of automation and reduced bias when\ninspecting the products. Furthermore, the increased speed of inspection reduces\noverall costs and time required for defect inspection. In this research, we\ncompare five streaming machine learning algorithms applied to visual defect\ninspection with real-world data provided by Philips Consumer Lifestyle BV.\nFurthermore, we compare them in a streaming active learning context, which\nreduces the data labeling effort in a real-world context. Our results show that\nactive learning reduces the data labeling effort by almost 15% on average for\nthe worst case, while keeping an acceptable classification performance. The use\nof machine learning models for automated visual inspection are expected to\nspeed up the quality inspection up to 40%.\n