| 摘要(英) |
From the actual factory experience of Automatic Optical Inspection System (AOI System) development, we know that defect data is hard to collect and need a lot of time to label, often makes project behind schedule. If we want to obtain enough data for training in the early stage of project, data augmentation technique is necessary. However, common data augmentation method like standard augmentation (flip, shift) and most of deep learning generative architecture do not consider that if the product data really exist in real world. Directly implement these method may produce data which do not meet domain knowledge. Therefore, we intend to develop a data augmentation system that can generate high quality images and products can be controlled with domain knowledge.
Pixel-wise generative guide are prepared to provide domain knowledge for generation process. Due to limited data, the implementation result of famous GAN model Pix2Pix (use generative guide image as input condition) was unfavorable. After paper researching, we get inspiration and propose Generative Layer Combiner (GLC). A generative system which combine rule-based method and GAN. GLC will separate each image in the dataset into three layers, three components. Three components from component libraries will be combined by rule-based image combiner which follows domain knowledge. After rule-based combination, boundary image of each layer will be refined by GAN-based deep learning modifier.
From the experiment results, we can see that compare to baseline model Pix2Pix, GLC can produce higher quality images and reduce 79% real data requirement when training deep learning model. Moreover, compare to no augmentation CNN classifier, CNN classifier training with GLC augmentation’s performance improve by reduce error rate with relative IMP of 97%. |
| 參考文獻 |
[1] Maayan Frid-Adar, Idit Diamant, Eyal Klang, Michal Amitai, Jacob Goldberger, Hayit Greenspan, “GAN-based Synthetic Medical Image Augmentation for increased CNN Performance in Liver Lesion Classification,” Neurocomputing
Volume 321, 2018, Pages 321-331.
[2] Shuanlong Niu , Bin Li, Xinggang Wang , Hui Lin, “Defect Image Sample Generation With GAN for Improving Defect Recognition,” IEEE Transactions on Automation Science and Engineering Volume: 17, 2020, Pages 1611 - 1622.
[3] Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, Alexei A. Efros, “Image-to-Image Translation with Conditional Adversarial Networks,” 2017 Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 5967 - 5976.
[4] Jianwei Yang, Anitha Kannan, Dhruv Batra, Devi Parikh, “LR-GAN: Layered Recursive Generative Adversarial Networks for Image Generation,” 2017 International Conference on Learning Representations (ICLR), 2017.
[5] Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Advances in Neural Information Processing Systems 25 (NIPS 2012), 2012.
[6] Karen Simonyan, Andrew Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” International Conference on Learning Representations (ICLR), 2015.
[7] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun, “Deep Residual Learning for Image Recognition,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
[8] Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, Yoshua Bengio,“Generative Adversarial Networks,” Proceedings of the International Conference on Neural Information Processing Systems (NIPS 2014), 2014, pp. 2672–2680.
[9] Xudong Mao, Qing Li, Haoran Xie, Raymond Y.K. Lau, Zhen Wang, Stephen Paul Smolley,“Least Squares Generative Adversarial Networks,”2017 IEEE International Conference on Computer Vision (ICCV), 2017, pp. 2813 - 2821.
[10] Mehdi Mirza, Simon Osindero, “Conditional Generative Adversarial Nets,” arXiv preprint arXiv:1411.1784, 2014.
[11] Jun-Yan Zhu, Taesung Park, Phillip Isola, Alexei A. Efros, “Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks,” 2017 IEEE International Conference on Computer Vision (ICCV), 2017, pp. 2242 - 2251.
[12] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, "Grad-cam: Visual explanations from deep networks via gradient-based localization," Proceedings of the IEEE international conference on computer vision, 2017, pp. 618-626. |
[Endnote RIS 格式]
[相關文章]
[文章引用]
[完整記錄]
[館藏目錄]
至系統瀏覽論文 ( 永不開放)
plurk
funp
live
udn
HD
myshare
netvibes
friend
youpush
delicious
baidu
Google bookmarks
del.icio.us
hemidemi
facebook
twitter
google
reddit