Dyomin Vladislav Dmitrievich (Post-graduate student,
Moscow City University (MCU), Moscow, Russia
)
Romashkova Oxana Nikolaevna (Doctor of Engineering, Professor,
Russian Presidential Academy of National Economy
and Public Administration (RANEPA), Moscow, Russia
)
| |
The paper addresses the problem of anomaly detection in visual data of robotic production systems. A hybrid method combining convolutional networks, a transformer module, and reconstruction-based analysis is proposed. The approach enables robust detection of both local and global defects and operates without anomaly labels. Experiments demonstrate improved accuracy and robustness to domain shifts compared to existing methods, confirming its applicability to industrial quality control.
Keywords:anomaly detection, machine vision, robotic systems, hybrid models, convolutional networks, transformers; reconstruction, quality control
|
|
| |
|
Read the full article …
|
Citation link:
Dyomin V. D., Romashkova O. N. A HYBRID DEEP LEARNING METHOD FOR ANOMALY DETECTION IN VISUAL DATA OF ROBOTIC PRODUCTION SYSTEMS // Современная наука: актуальные проблемы теории и практики. Серия: Естественные и Технические Науки. -2026. -№01. -С. 78-84 DOI 10.37882/2223-2966.2026.01.13 |
|
|
LEGAL
INFORMATION:
