Related Experiment Video
Updated: Jan 19, 2026
05:47
Simulation of a Scaled Assembly Process with Collaboration of a Robotic Arm and Monitoring through a Vision System for Quality Control
Published on: August 29, 2025
426
Deep Learning for Industrial Computer Vision Quality Control in the Printing Industry 4.0
Javier Villalba-Diez1,2, Daniel Schmidt3,4, Roman Gevers5
1Hochschule Heilbronn, Fakultät Management und Vertrieb, Campus Schwäbisch Hall, 74523 Schwäbisch Hall, Germany. javier.villalba-diez@hs-heilbronn.de.
Sensors (Basel, Switzerland)
|September 22, 2019
Summary
This study introduces a deep neural network (DNN) soft sensor for automated industrial visual inspection in the printing industry. The DNN system enhances defect detection accuracy and reduces costs by comparing scanned cylinder surfaces to engraving files.
Area of Science:
- Manufacturing Technology
- Artificial Intelligence
- Quality Control
Background:
- Industrial inspection faces challenges in balancing accuracy, speed, and cost.
- Automating visual inspection processes is crucial for Industry 4.0.
- Defects in gravure cylinder production necessitate efficient quality control.
Purpose of the Study:
- To develop a deep neural network (DNN) soft sensor for automated quality control in gravure cylinder production.
- To enhance defect detection accuracy and reduce inspection costs through process automation.
- To integrate a DNN soft sensor with high-resolution optical cameras for industrial visual inspection.
Main Methods:
- A deep neural network (DNN) was developed as a soft sensor.
- The DNN compares scanned cylinder surfaces against the original engraving file.
- The system learns features from training data to perform automatic quality control.
Main Results:
- The developed DNN sensor achieved high accuracy in defect detection (specific metric missing).
- The application demonstrated potential for cost reduction in industrial visual inspection.
- The system successfully automated a part of the quality control process.
Conclusions:
- The DNN soft sensor offers a promising approach to improve industrial visual inspection accuracy and efficiency.
- Future work will focus on predicting error amounts, supporting human operators, and enabling autonomous quality decisions.
- This technology contributes to the advancement of automated quality control in the printing industry.
