| Autor: |
Rožanec, Jože M., Zajec, Patrik, Keizer, Jelle, Trajkova, Elena, Fortuna, Blaž, Brecelj, Bor, Šircelj, Beno, Mladeni´c, Dunja |
| Předmět: |
|
| Zdroj: |
Central European Conference on Information & Intelligent Systems; 2022, p357-363, 7p |
| Abstrakt: |
Quality control allows companies to verify whether the products conform to requirements and specifications. However, while Artificial Intelligence is increasingly used to automate the visual inspection process, a manual revision can be required when the model cannot determine whether a piece is defective or not with enough confidence. Therefore, means must be devised to optimize the manual revision of such products, to increase the speed and quality of labeling. In this paper, we perform experiments to determine whether different defect hinting techniques and data imbalance mitigation techniques can enhance the manual revision process. Furthermore, we contrast the performance of two groups of persons with different skills and education levels and their perceptions when executing the experiments. We performed the experiments on real-world data provided by Philips Consumer Lifestyle BV. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
