Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Synthetic data generation framework for quality control automation in gravure printing

Source

arXiv

arxiv.org/abs/2607.21577v1

Read the source

Source

Plain-Language Brief

Researchers have developed a synthetic data generation framework to help automate the quality control process in gravure printing. The framework generates high-fidelity images of common printing defects and uses them to train deep learning models for object detection. An experiment shows that the state-of-the-art RFDETR model, trained on the synthetic data, achieves an 80.9% Mean Average Precision (mAP) when tested on real industrial samples.

What the Document Says

  • The research is focused on improving quality control in gravure printing through automation of defect detection.
  • Deep learning models, such as YOLO and Vision Transformers, are mentioned as potential solutions for automation but are hindered by the scarcity of real-world industrial defect images.
  • A novel synthetic data generation framework is proposed to generate high-fidelity images of specific printing defects for training deep learning models.
  • The framework produces bounding boxes and annotations along with the generated images.
  • To validate the framework, a synthetic dataset of 7533 images was created and used to train RFDETR, a state-of-the-art object detection model.
  • The model trained on the synthetic data achieved a Mean Average Precision (mAP) of 80.9% when tested on real industrial samples.
  • The proposed framework offers a cost-effective and rapid deployment solution for automating defect inspection in printing lines.

Key Concepts Mentioned

  • quality control
  • gravure printing
  • deep learning
  • synthetic data generation
  • RFDETR
  • Mean Average Precision (mAP)

Claims Made by the Document

  • Deep learning models can help automate quality control in gravure printing.
  • Scarcity of real-world industrial defect images hinders training robust deep learning models for gravure printing.
  • The proposed synthetic data generation framework can generate high-fidelity images of specific printing defects for training deep learning models.
  • An experiment shows that the RFDETR model, trained on the synthetic data, achieves a Mean Average Precision (mAP) of 80.9% when tested on real industrial samples.

Evidence Used

  • The text mentions that deep learning models are potential solutions for automation but are hindered by the scarcity of real-world industrial defect images.
  • The proposed synthetic data generation framework is described and its use for generating high-fidelity images of printing defects is detailed.
  • A synthetic dataset of 7533 images was created and used to train RFDETR, with experimental results showing an 80.9% Mean Average Precision (mAP).

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Quality control in printing, particularly in rotogravure printing, still depends on slow, costly, and subjective manual inspection. Automated surface defect detection is critical for maintaining high-quality standards in rotogravure printing. Deep learning models give prospects for automation. However, training robust deep learning models, such as YOLO or Vision Transformers, is heavily hindered by the extreme scarcity of real-world industrial defects images. To overcome this limitation, this paper introduces a novel synthetic data generation framework tailored for rotogravure printing quality control. The proposed pipeline automatically generates high-fidelity images of specific printing defects (creases, streaks, misregistration, etc.) and outputs corresponding bounding boxes and annotations. To validate the framework, a synthetic dataset of 7533 images was generated and used to train the state-of-the-art object-detection model RFDETR. Experimental results demonstrate that the model trained on our synthetic data achieves a Mean Average Precision (mAP) of 80.9% on real industrial testing samples. This framework provides a zero-cost, rapid-deployment solution for automating defect inspection in printing lines without requiring massive manual data collection.