Data Augmentation for Semantic Segmentation in the Context of Carbon Fiber Defect Detection using Adversarial Learning

Mertes, Silvan; Margraf, Andreas; Kommer, Christoph; Geinitz, Steffen; André, Elisabeth

doi:10.5220/0009823500590067

Cited by 10 publications

(20 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also, it is conceivable to use the proposed system for the task of data augmentation. Previous work has shown that GANs in general have the ability to enhance datasets in order to improve various deep learning tasks, such as semantic segmentation of images (Scherer et al, 2021;Choi et al, 2019;Mertes et al, 2020b;Uricar et al, 2019) or various imageand audio-based classification problems (Mariani et al, 2018;Frid-Adar et al, 2018;Waheed et al, 2020;Mertes et al, 2020a). The ability to abstract from discrete classes to continuous features opens up a variety of machine learning problems where label interpolation could improve performance through data augmentation, which we plan to study in further research.…”

Section: Discussionmentioning

confidence: 99%

Intercategorical Label Interpolation for Emotional Face Generation with Conditional Generative Adversarial Networks

Mertes¹,

Schiller²,

Lingenfelser³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Generative adversarial networks offer the possibility to generate deceptively real images that are almost indistinguishable from actual photographs. Such systems however rely on the presence of large datasets to realistically replicate the corresponding domain. This is especially a problem if not only random new images are to be generated, but specific (continuous) features are to be co-modeled. A particularly important use case in Human-Computer Interaction (HCI) research is the generation of emotional images of human faces, which can be used for various use cases, such as the automatic generation of avatars. The problem hereby lies in the availability of training data. Most suitable datasets for this task rely on categorical emotion models and therefore feature only discrete annotation labels. This greatly hinders the learning and modeling of smooth transitions between displayed affective states. To overcome this challenge, we explore the potential of label interpolation to enhance networks trained on categorical datasets with the ability to generate images conditioned on continuous features.

show abstract

Section: Discussionmentioning

confidence: 99%

Intercategorical Label Interpolation for Emotional Face Generation with Conditional Generative Adversarial Networks

Mertes¹,

Schiller²,

Lingenfelser³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…The approach allows for artificially creating label and image pairs that are actually new and serve as training samples for semantic segmentation models. For the generation of new labels, we propose two distinct concepts: the first approach is based on a function precisely tailored to the application and was already introduced by Mertes et al (Mertes. et al, 2020b).…”

Section: Pushing the Limits Of Image Segmentationmentioning

confidence: 99%

“…To create new label data, we propose two different methods. The first method, that we already introduced in (Mertes. et al, 2020b), is an algorithm specifically designed for our particular application at hand, i.e.…”

Section: Approachmentioning

confidence: 99%

See 1 more Smart Citation

Alternative Data Augmentation for Industrial Monitoring using Adversarial Learning

Mertes¹,

Margraf²,

Geinitz³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Visual inspection software has become a key factor in the manufacturing industry for quality control and process monitoring. Semantic segmentation models have gained importance since they allow for more precise examination. These models, however, require large image datasets in order to achieve a fair accuracy level. In some cases, training data is sparse or lacks of sufficient annotation, a fact that especially applies to highly specialized production environments. Data augmentation represents a common strategy to extend the dataset. Still, it only varies the image within a narrow range. In this article, a novel strategy is proposed to augment small image datasets. The approach is applied to surface monitoring of carbon fibers, a specific industry use case. We apply two different methods to create binary labels: a problem-tailored trigonometric function and a WGAN model. Afterwards, the labels are translated into color images using pix2pix and used to train a U-Net. The results suggest that the trigonometric function is superior to the WGAN model. However, a precise examination of the resulting images indicate that WGAN and image-to-image translation achieve good segmentation results and only deviate to a small degree from traditional data augmentation. In summary, this study examines an industry application of data synthesization using generative adversarial networks and explores its potential for monitoring systems of production environments.

show abstract

“…this means that the network needs to be trained from scratch each time a change in the target properties of the data is required. Artificial data that was generated by GANs has been used for data augmentation predominantly in the field of image processing (e.g., [9]- [11]), but there is also recent work that makes use of GAN-based data augmentation for acoustic scene classification [12]- [14] as well as emotional speech [15], [16]. Most of these approaches are not able to generate the augmented data in a controlled manner, but rather use the GANs to produce random new samples to enhance existing datasets.…”

Section: Related Workmentioning

confidence: 99%

An Evolutionary-based Generative Approach for Audio Data Augmentation

Mertes

Baird

Schiller

et al. 2020

2020 IEEE 22nd International Workshop on Multimedia Signal Processing (MMSP)

Self Cite

View full text Add to dashboard Cite

In this paper, we introduce a novel framework to augment raw audio data for machine learning classification tasks. For the first part of our framework, we employ a generative adversarial network (GAN) to create new variants of the audio samples that are already existing in our source dataset for the classification task. In the second step, we then utilize an evolutionary algorithm to search the input domain space of the previously trained GAN, with respect to predefined characteristics of the generated audio. This way we are able to generate audio in a controlled manner that contributes to an improvement in classification performance of the original task. To validate our approach, we chose to test it on the task of soundscape classification. We show that our approach leads to a substantial improvement in classification results when compared to a training routine without data augmentation and training with uncontrolled data augmentation with GANs.

show abstract

Data Augmentation for Semantic Segmentation in the Context of Carbon Fiber Defect Detection using Adversarial Learning

Cited by 10 publications

References 15 publications

Intercategorical Label Interpolation for Emotional Face Generation with Conditional Generative Adversarial Networks

Intercategorical Label Interpolation for Emotional Face Generation with Conditional Generative Adversarial Networks

Alternative Data Augmentation for Industrial Monitoring using Adversarial Learning

An Evolutionary-based Generative Approach for Audio Data Augmentation

Contact Info

Product

Resources

About