Development of a Convolutional Layer of a Neural Network for Detecting Defects in Sheet Metal Products on Defectoscopic Images
- Resource Type
- Conference
- Authors
- Mortin, Constantin V.; Privezentsev, Denis G.
- Source
- 2021 XXIV International Conference on Soft Computing and Measurements (SCM) Soft Computing and Measurements (SCM),2021 XXIV International Conference on. :154-156 May, 2021
- Subject
- Bioengineering
Communication, Networking and Broadcast Technologies
Computing and Processing
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Deep learning
Geometry
Image segmentation
Coordinate measuring machines
Filtering
Metals
Production
flaw detection image
rolled metal defect
technical vision
neural networks
- Language
The main problems of timely detection of defects on sheet metal by means of technical vision are considered. In the course of the analysis, it was found that artificial neural networks of a typical structure do not allow to reduce the influence of real production factors on digital flaw detection images, and the quality of defect detection will be quite high. A neural network of a special structure has been created and specialized algorithms have been developed based on the created network.