Journal of the Korea Institute of Information and Communication Engineering (한국정보통신학회논문지)

The Korea Institute of Information and Commucation Engineering (한국정보통신학회)
권호 표지
DOI QR코드

DOI QR Code

A Broken Image Screening Method based on Histogram Analysis to Improve GAN Algorithm

GAN 알고리즘 개선을 위한 히스토그램 분석 기반 파손 영상 선별 방법

  • Cho, Jin-Hwan (Department of Software Convergence, Dong-Eui University) ;
  • Jang, Jongwook (Department of Computer Engineering, Dong-Eui University) ;
  • Jang, Si-Woong (Department of Computer Engineering, Dong-Eui University)
  • Received : 2021.10.19
  • Accepted : 2021.10.30
  • Published : 2022.04.30
Download PDF
⟨ Previous Next ⟩

Abstract

Recently, many studies have been done on the data augmentation technique as a way to efficiently build datasets. Among them, a representative data augmentation technique is a method of utilizing Generative Adversarial Network (GAN), which generates data similar to real data by competitively learning generators and discriminators. However, when learning GAN, there are cases where a broken pixel image occurs among similar data generated according to the environment and progress, which cannot be used as a dataset and causes an increase in learning time. In this paper, an algorithm was developed to select these damaged images by analyzing the histogram of image data generated during the GAN learning process, and as a result of comparing them with the images generated in the existing GAN, the ratio of the damaged images was reduced by 33.3 times(3,330%).

최근 데이터셋을 효율적으로 구축하는 방법으로 데이터 증강 기법과 관련하여 많은 연구가 이루어지고 있다. 이 중 대표적인 데이터 증강 기법은 생성적 적대 신경망(Generative Adversarial Network:GAN)을 활용하는 방법이며, 이는 생성자와 판별자를 서로 경쟁 학습시킴으로써 진짜 데이터와 유사한 데이터를 생성해내는 기법이다. 그러나, GAN을 학습할 때 환경 및 진행 정도에 따라 생성되는 유사 데이터 중에서 픽셀이 깨지는 파손 영상이 발생하는 경우가 있으며, 이러한 영상은 데이터셋으로 활용할 수 없고 학습 시간을 증가시키는 원인이 된다. 본 논문에서는 GAN 학습 과정에서 생성되는 영상 데이터의 히스토그램을 분석하여 이러한 파손 영상을 선별해내는 알고리즘을 개발하였으며, 기존 GAN에서 생성되는 영상과 비교해 본 결과 파손 영상의 비율을 33.3배(3,330%) 감소시켰다.

Keywords

Acknowledgement

This research was supported by the MSIT (Ministry of Science and ICT),Korea, under the Grand Information Technology Research Center support program(IITP-2022-2020-0-01791) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation)

References

  1. K. M. Choi, Y. M. Kim, J. P. Shin, S. M. Sung, and B. K. Lee, "Data set design and implementation for Assistive walking device AI service construction," in Proceedings of the Korean Society of Computer Information Conference, Busan, vol. 29, no. 1, pp. 227-229, Jan. 2021.
  2. J. S. Lee, B. K. Ko, E. S. Kang, H. J. Choi, J. O. Kim, and B. K. Lee, "AI Learning Cookie Image Data Set Construction," in Proceedings of the Korean Society of Computer Information Conference, Jeju, vol. 28, no. 2, pp. 347-349, Jul. 2020.
  3. H. C. Lee, S. Y. Shin, "Development of Data Collection System using Google Environment," in Proceedings of The Korean Institute of Information and Communication Science, Busan, vol. 23, no. 2, pp. 704-705, Oct. 2019.
  4. J. S. Kim and S. W. Jang, "Construction Method of Multifaceted Image Datasets for Improving Object Recognition Rate in Deep Learniung System," in International Conference on Future Information & Communication Engineering, Online, vol. 12, no. 1, pp. 144-147, Feb. 2021.
  5. E. S. Park, Y. J. Yang, J. H. Jeon, and E. S. Ryu, "Image Web Crawling Program for Artificial Intelligence Datasets," in Proceedings of the Korean Society of Broad Engineers, Seoul, p. 55, Nov. 2018.
  6. J. H. Choi, K. M. Irick, J. Hardin, W. Qiu, A. Yuille, J. Sampson, and V. Narayanan, "Stochastic Functional Verification of DNN Design through Progressive Virtual Dataset Generation," in 2018 IEEE International Symposiumon Circuits and Systems(ISCAS), Florence, pp. 1-5, 2018.
  7. I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, "Generative Adversarial Networks" arXiv: 1406.2661 [stat.ML], Jun. 2014. Available: https://arxiv.org/abs/1406.2661
  8. S. H. Lim, Y. G. Shin, C. H. Yoo, H. K. Lee, and S. J. Ko, "Data Augmentation method using WGAN," in 2017 Institute of Electronics and Information Engineers Fall Conference(IEIE), Incheon, pp. 516-519, Nov. 2017.
  9. Y. J. Yang, Y. G. Hong, and J. H. Park, "Efficient Learning Dataset Generation and Data Selection Using Generative Adversarial Network and GSVD-Based Linear Discriminant Analysis," The Journal of Korean Institute of Communications and Information Sciences, vol. 45, no. 7, pp. 1166-1173, Jul. 2020. https://doi.org/10.7840/kics.2020.45.7.1166
  10. M. Onder and Y. S. Akgul, "Automatic Generation of Matching Clothes Design Using Generative Adversarial Networks," in 2020 28th Signal Processing and Communications Applications Conference (SIU), Gaziantep, pp. 1-4, 2020.
  11. J. H. Jeong, J. W. Kim, and H. T. Kim, "Expanding Training Datasets for Image Classification Network using GAN," in Proceedings of The Korean Institute of Information and Communication Science, vol. 23, no. 2, pp. 75-76, Oct. 2019.
  12. S. J. Bae, M. G. Kim, and H. K. Jung, "GAN System Using Noise for Image Generation," Journal of the Korea Institute of Information and Communication Engineering, vol. 24, no. 6, pp. 700-705, Jun. 2020. https://doi.org/10.6109/JKIICE.2020.24.6.700
  13. T. Kerras, S. Laine, and T. Aila, "A Style-Based Generator Architecture for Generative Adversarial Networks," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), CA: California, pp. 4401-4410, 2019.
  14. T. Kerras, S. Laine, M. Aittala, J. Hellsten, J. Lehtinen, and T. Aila, "Analyzing and Improving the Image Quality of StyleGAN," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), WA: Washington, pp. 8110-8119, 2020.
  15. NVlabs / StyleGAN2 - Training Networks [Internet]. Available: https://github.com/NVlabs/stylegan2