Integration of AI-Based Automatic Defect Recognition (ADR) in X-ray Inspection of Castings
e-Journal of Nondestructive Testing · 2026-07 · 31 (7)
DOI: 10.58286/33062 ↗Kurz gefasst
Fachartikel in der Zeitschrift e-Journal of Nondestructive Testing, 2026-07. Verfasst von Thomas Stocker und 4 weiteren Autor:innen.
Abstract (Originalsprache)
AI-based automatic defect recognition (ADR) in X-ray images of castings offers significant potential for improving inspection efficiency by reducing setup time, streamlining decision-making processes, and minimizing manual verification efforts. However, customers face substantial entry barriers when adopting this technology. This paper presents a practical solution developed by Fraunhofer IIS that addresses these challenges. The approach combines a pre-trained neural network based on both simulated and real X-ray data with easy integration into the established ISAR inspection system. Customers require only a small number of their own images to achieve high inspection performance. The system enables seamless integration into existing and new installations through an optional module within ISAR. A performance comparison with classical ADR is presented for two industrial applications: light metal wheels and chassis components. Results demonstrate near-100% probability of detection (POD) for critical defects while maintaining low false positive rates, significantly reducing manual verification efforts.
Zitieren
APA
Stocker, T., Gosswami, B. M., Chanda, R. K., Sukowski, F., & Schön, T. (2026). Integration of AI-Based Automatic Defect Recognition (ADR) in X-ray Inspection of Castings. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/33062
Vancouver
Stocker T, Gosswami BM, Chanda RK, Sukowski F, Schön T. Integration of AI-Based Automatic Defect Recognition (ADR) in X-ray Inspection of Castings. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/33062
BibTeX
@article{Stocker2026,
author = {Stocker, Thomas and Gosswami, Bishwajit Mohan and Chanda, Rajib Kumar and Sukowski, Frank and Schön, Tobias},
title = {Integration of AI-Based Automatic Defect Recognition (ADR) in X-ray Inspection of Castings},
journal = {e-Journal of Nondestructive Testing},
year = {2026},
volume = {31},
number = {7},
doi = {10.58286/33062}
}
EUVIMED