On-the-fly image-level oversampling for imbalanced datasets of manufacturing defects.
Saved in:
| Title: | On-the-fly image-level oversampling for imbalanced datasets of manufacturing defects. |
|---|---|
| Authors: | Theodoropoulos, Spyros1,2 (AUTHOR) stheodoropoulos@mail.ntua.gr, Zajec, Patrik3 (AUTHOR), Rožanec, Jože M.3 (AUTHOR), Kyriazis, Dimosthenis2 (AUTHOR), Tsanakas, Panayiotis1 (AUTHOR) |
| Source: | Machine Learning. Jul2024, Vol. 113 Issue 7, p4013-4035. 23p. |
| Subjects: | Data augmentation, Class differences, Generative adversarial networks, Computational neuroscience |
| Abstract: | Visual defect recognition and its manufacturing applications have been an upcoming topic in recent AI research. Defect datasets are often severely imbalanced and can be additionally burdened with separating classes of high visual similarity. Although various methods of data augmentation have been proposed to mitigate the class imbalance, they often fail to cope with tinier minority classes or have fidelity issues with smaller defects while, at the same time, needing significant computational resources to train. Also, augmentation based on vector-based oversampling struggles to produce high-fidelity inputs and is hard to apply on custom CNN architectures, which often perform better for this type of problem. Our work presents an image-level oversampling method based on an instance-based image generator that can be applied to any CNN directly during the training process without increasing the order of training time required. It is based on identifying a small number of the most uncertain base samples close to the estimated class boundaries and using them as seeds for augmentation. The resulting images are of high visual quality preserving small class differences, and they also improve the classifier boundary leading to higher recall scores than other state-of-the-art approaches. [ABSTRACT FROM AUTHOR] |
| Copyright of Machine Learning is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| FullText | Links: – Type: pdflink Text: Availability: 1 |
|---|---|
| Header | DbId: egs DbLabel: Engineering Source An: 177598360 AccessLevel: 6 PubType: Academic Journal PubTypeId: academicJournal PreciseRelevancyScore: 0 |
| IllustrationInfo | |
| Items | – Name: Title Label: Title Group: Ti Data: On-the-fly image-level oversampling for imbalanced datasets of manufacturing defects. – Name: Author Label: Authors Group: Au Data: <searchLink fieldCode="AR" term="%22Theodoropoulos%2C+Spyros%22">Theodoropoulos, Spyros</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<i> stheodoropoulos@mail.ntua.gr</i><br /><searchLink fieldCode="AR" term="%22Zajec%2C+Patrik%22">Zajec, Patrik</searchLink><relatesTo>3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Rožanec%2C+Jože+M%2E%22">Rožanec, Jože M.</searchLink><relatesTo>3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Kyriazis%2C+Dimosthenis%22">Kyriazis, Dimosthenis</searchLink><relatesTo>2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Tsanakas%2C+Panayiotis%22">Tsanakas, Panayiotis</searchLink><relatesTo>1</relatesTo> (AUTHOR) – Name: TitleSource Label: Source Group: Src Data: <searchLink fieldCode="JN" term="%22Machine+Learning%22">Machine Learning</searchLink>. Jul2024, Vol. 113 Issue 7, p4013-4035. 23p. – Name: Subject Label: Subjects Group: Su Data: <searchLink fieldCode="DE" term="%22Data+augmentation%22">Data augmentation</searchLink><br /><searchLink fieldCode="DE" term="%22Class+differences%22">Class differences</searchLink><br /><searchLink fieldCode="DE" term="%22Generative+adversarial+networks%22">Generative adversarial networks</searchLink><br /><searchLink fieldCode="DE" term="%22Computational+neuroscience%22">Computational neuroscience</searchLink> – Name: Abstract Label: Abstract Group: Ab Data: Visual defect recognition and its manufacturing applications have been an upcoming topic in recent AI research. Defect datasets are often severely imbalanced and can be additionally burdened with separating classes of high visual similarity. Although various methods of data augmentation have been proposed to mitigate the class imbalance, they often fail to cope with tinier minority classes or have fidelity issues with smaller defects while, at the same time, needing significant computational resources to train. Also, augmentation based on vector-based oversampling struggles to produce high-fidelity inputs and is hard to apply on custom CNN architectures, which often perform better for this type of problem. Our work presents an image-level oversampling method based on an instance-based image generator that can be applied to any CNN directly during the training process without increasing the order of training time required. It is based on identifying a small number of the most uncertain base samples close to the estimated class boundaries and using them as seeds for augmentation. The resulting images are of high visual quality preserving small class differences, and they also improve the classifier boundary leading to higher recall scores than other state-of-the-art approaches. [ABSTRACT FROM AUTHOR] – Name: AbstractSuppliedCopyright Label: Group: Ab Data: <i>Copyright of Machine Learning is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.) |
| PLink | https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=177598360 |
| RecordInfo | BibRecord: BibEntity: Identifiers: – Type: doi Value: 10.1007/s10994-023-06498-4 Languages: – Code: eng Text: English PhysicalDescription: Pagination: PageCount: 23 StartPage: 4013 Subjects: – SubjectFull: Data augmentation Type: general – SubjectFull: Class differences Type: general – SubjectFull: Generative adversarial networks Type: general – SubjectFull: Computational neuroscience Type: general Titles: – TitleFull: On-the-fly image-level oversampling for imbalanced datasets of manufacturing defects. Type: main BibRelationships: HasContributorRelationships: – PersonEntity: Name: NameFull: Theodoropoulos, Spyros – PersonEntity: Name: NameFull: Zajec, Patrik – PersonEntity: Name: NameFull: Rožanec, Jože M. – PersonEntity: Name: NameFull: Kyriazis, Dimosthenis – PersonEntity: Name: NameFull: Tsanakas, Panayiotis IsPartOfRelationships: – BibEntity: Dates: – D: 01 M: 07 Text: Jul2024 Type: published Y: 2024 Identifiers: – Type: issn-print Value: 08856125 Numbering: – Type: volume Value: 113 – Type: issue Value: 7 Titles: – TitleFull: Machine Learning Type: main |
| ResultId | 1 |
