Rail surface defect data enhancement method based on improved ACGAN
Rail surface defects present a significant safety concern in railway operations. However, the scarcity of data poses challenges for employing deep learning in defect detection. This study proposes an enhanced ACGAN augmentation method to address these issues. Residual blocks mitigate vanishing gradi...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2024-04-01
|
| Series: | Frontiers in Neurorobotics |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fnbot.2024.1397369/full |
| _version_ | 1797216424121335808 |
|---|---|
| author | He Zhendong Gao Xiangyang Liu Zhiyuan An Xiaoyu Zheng Anping |
| author_facet | He Zhendong Gao Xiangyang Liu Zhiyuan An Xiaoyu Zheng Anping |
| author_sort | He Zhendong |
| collection | DOAJ |
| description | Rail surface defects present a significant safety concern in railway operations. However, the scarcity of data poses challenges for employing deep learning in defect detection. This study proposes an enhanced ACGAN augmentation method to address these issues. Residual blocks mitigate vanishing gradient problems, while a spectral norm regularization-constrained discriminator improves stability and image quality. Substituting the generator’s deconvolution layer with upsampling and convolution operations enhances computational efficiency. A gradient penalty mechanism based on regret values addresses gradient abnormality concerns. Experimental validation demonstrates superior image clarity and classification accuracy compared to ACGAN, with a 17.6% reduction in FID value. MNIST dataset experiments verify the model’s generalization ability. This approach offers practical value for real-world applications. |
| first_indexed | 2024-04-24T11:45:44Z |
| format | Article |
| id | doaj.art-46e2b874b91641deabf6e50da14e11df |
| institution | Directory Open Access Journal |
| issn | 1662-5218 |
| language | English |
| last_indexed | 2024-04-24T11:45:44Z |
| publishDate | 2024-04-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neurorobotics |
| spelling | doaj.art-46e2b874b91641deabf6e50da14e11df2024-04-09T10:23:24ZengFrontiers Media S.A.Frontiers in Neurorobotics1662-52182024-04-011810.3389/fnbot.2024.13973691397369Rail surface defect data enhancement method based on improved ACGANHe Zhendong0Gao Xiangyang1Liu Zhiyuan2An Xiaoyu3Zheng Anping4School of Electrical and Information Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaSchool of Electrical and Information Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaSchool of Rail Transit Engineering, Zhengzhou Technical College, Zhengzhou, ChinaSchool of Electrical and Information Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaSchool of Electrical and Information Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaRail surface defects present a significant safety concern in railway operations. However, the scarcity of data poses challenges for employing deep learning in defect detection. This study proposes an enhanced ACGAN augmentation method to address these issues. Residual blocks mitigate vanishing gradient problems, while a spectral norm regularization-constrained discriminator improves stability and image quality. Substituting the generator’s deconvolution layer with upsampling and convolution operations enhances computational efficiency. A gradient penalty mechanism based on regret values addresses gradient abnormality concerns. Experimental validation demonstrates superior image clarity and classification accuracy compared to ACGAN, with a 17.6% reduction in FID value. MNIST dataset experiments verify the model’s generalization ability. This approach offers practical value for real-world applications.https://www.frontiersin.org/articles/10.3389/fnbot.2024.1397369/fullACGANdata enhancementresidual blockspectral norm regularizationgradient punishment |
| spellingShingle | He Zhendong Gao Xiangyang Liu Zhiyuan An Xiaoyu Zheng Anping Rail surface defect data enhancement method based on improved ACGAN Frontiers in Neurorobotics ACGAN data enhancement residual block spectral norm regularization gradient punishment |
| title | Rail surface defect data enhancement method based on improved ACGAN |
| title_full | Rail surface defect data enhancement method based on improved ACGAN |
| title_fullStr | Rail surface defect data enhancement method based on improved ACGAN |
| title_full_unstemmed | Rail surface defect data enhancement method based on improved ACGAN |
| title_short | Rail surface defect data enhancement method based on improved ACGAN |
| title_sort | rail surface defect data enhancement method based on improved acgan |
| topic | ACGAN data enhancement residual block spectral norm regularization gradient punishment |
| url | https://www.frontiersin.org/articles/10.3389/fnbot.2024.1397369/full |
| work_keys_str_mv | AT hezhendong railsurfacedefectdataenhancementmethodbasedonimprovedacgan AT gaoxiangyang railsurfacedefectdataenhancementmethodbasedonimprovedacgan AT liuzhiyuan railsurfacedefectdataenhancementmethodbasedonimprovedacgan AT anxiaoyu railsurfacedefectdataenhancementmethodbasedonimprovedacgan AT zhenganping railsurfacedefectdataenhancementmethodbasedonimprovedacgan |