Summary of Machine Learning in Industrial Quality Control Of Glass Bottle Prints, by Maximilian Bundscherer et al.
Machine Learning in Industrial Quality Control of Glass Bottle Prints
by Maximilian Bundscherer, Thomas H. Schmitt, Tobias Bocklet
First submitted to arxiv on: 30 Sep 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A machine learning-based quality control system was developed for industrial glass bottle printing processes. The system can detect minor defects despite reflections or manufacturing-related deviations. Two approaches were evaluated: one using filters to suppress reflections and image quality metrics as features for supervised classification models, achieving an accuracy of 84%. Another approach fine-tuned pre-trained CNN models for binary classification, resulting in an accuracy of 87%. Grad-Cam was used to visualize frequently defective bottle print regions, providing insights for process optimization. The paper describes the general approach and challenges encountered during data collection, including unsupervised preselection and labeling. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way is being developed to check glass bottles for small defects before they are made. This helps make sure the bottles turn out right despite things that can go wrong like reflections or tiny changes in how they’re made. Two different methods were tested: one that uses filters to get rid of the reflections and measures how good an image is, which helps a computer learn what’s normal and what’s not. The other method takes pre-made learning models and makes them better at telling the difference between normal and abnormal bottles. Both ways were able to correctly identify problems about 85% of the time. This new way could help make glass bottles even better. |
Keywords
» Artificial intelligence » Classification » Cnn » Machine learning » Optimization » Supervised » Unsupervised
