DOI QR코드

DOI QR Code

A Study on the Complementary Method of Aerial Image Learning Dataset Using Cycle Generative Adversarial Network

CycleGAN을 활용한 항공영상 학습 데이터 셋 보완 기법에 관한 연구

  • Received : 2020.09.09
  • Accepted : 2020.11.13
  • Published : 2020.12.31

Abstract

This study explores how to build object classification learning data based on artificial intelligence. The data has been investigated recently in image classification fields and, in turn, has a great potential to use. In order to recognize and extract relatively accurate objects using artificial intelligence, a large amount of learning data is required to be used in artificial intelligence algorithms. However, currently, there are not enough datasets for object recognition learning to share and utilize. In addition, generating data requires long hours of work, high expenses and labor. Therefore, in the present study, a small amount of initial aerial image learning data was used in the GAN (Generative Adversarial Network)-based generator network in order to establish image learning data. Moreover, the experiment also evaluated its quality in order to utilize additional learning datasets. The method of oversampling learning data using GAN can complement the amount of learning data, which have a crucial influence on deep learning data. As a result, this method is expected to be effective particularly with insufficient initial datasets.

본 연구에서는 최근 영상판독 분야에서 활발히 연구되고, 활용성이 발전하고 있는 인공지능 기반 객체분류 학습 데이터 구축에 관한 내용을 다룬다. 영상판독분야에서 인공지능을 활용하여 정확도 높은 객체를 인식, 추출하기 위해서는 알고리즘에 적용할 많은 양의 학습데이터가 필수적으로 요구된다. 하지만, 현재 공동활용 가능한 데이터 셋이 부족할 뿐만 아니라 데이터 생성을 위해서는 많은 시간과 인력 및 고비용을 필요로 하는 것이 현실이다. 따라서 본 연구에서는 소량의 초기 항공영상 학습데이터를 GAN (Generative Adversarial Network) 기반의 생성기 신경망을 활용하여 오버샘플 영상 학습데이터를 구축하고, 품질을 평가함으로써 추가적 학습 데이터 셋으로 활용하기 위한 실험을 진행하였다. GAN을 이용하여 오버샘플 학습데이터를 생성하는 기법은 딥러닝 성능에 매우 중요한 영향을 미치는 학습데이터의 양을 획기적으로 보완할 수 있으므로 초기 데이터가 부족한 경우에 효과적으로 활용될 수 있을 것으로 기대한다.

Keywords

References

  1. Choi, H.W. and Seo, Y.C. (2020), Construction of synthesis training data using generative adversarial network, The Korea Association of Geographic Information Science Spring Conference-2020, pp.101-105. (in Korean)
  2. Isola, P., Zhu, J.Y., Zhou, T., and Alexei, A.E. (2018), Image-to-image translation with conditional adversarial networks, arXiv preprintarXiv: 1611.07004.
  3. Goodfellow, I., Pouget, J.A., Mirza, M., Xu, B., Warde D.F., Ozair, S., Courville, A., and Bengia, Y. (2014), Generative adversarial nets, Advances in Neural Information Processing Systems, pp. 2672-2680.
  4. Jeong, J.S. and Kim, Y.J. (2014), Structural similarity index for image assessment using pixel difference and saturation awareness, Journal of Korea Institute of Information Scientists and Engineers, Vol. 41, No. 10, pp. 847-858. (in Korean with English abstract)
  5. Kang, M.S., Park, S.W., and Yoon, K.S. (2006), Land cover classification of Image data using artificial neural networks, Journal of Korean Society of Rural Planning, Vol. 12, No. 1, pp. 75-83. (in Korean with English abstract)
  6. Kang, S.M. and Lee, J.J. (2019), Depth map extraction from the single image using Pix2Pix model, Journal of Korea Multimedia Society, Vol. 22, No. 5, pp. 547-557. (in Korean with English abstract) https://doi.org/10.9717/KMMS.2019.22.5.547
  7. Kim, H.O. and Yeom, J.M. (2012), A study on object-based image analysis methods for land cover classification in agricultural area, Journal of Korean Association of Geographic Information Studies, Vol. 15, No. 4, pp. 26-41. (in Korean with English abstract) https://doi.org/10.11108/kagis.2012.15.4.026
  8. Kim. K.H. and Park, S.S.(2008), A comparative study on broadcasting video quality using PSNR in IPTV network adopted transition mechanism, Journal of Korea.Institute of Maritime Information & Communication Sciences, Vol. 14, No. 1, pp. 156-166. (in Korean with English abstract)
  9. Lee, S.H. and Kim, J.S. (2019), Land cover classification using sematic image segmentation with deep learning, Korean Journal of Remote Sensing, Vol. 35, No. 2, pp. 279-288. (in Korean with English abstract) https://doi.org/10.7780/KJRS.2019.35.2.7
  10. Oh, C.H., Park, S.Y., Kim, H.S., Lee, Y.W., and Choi, C.U. (2010), Comparison of landcover map accuracy using high resolution satellite imagery, Journal of Korean Association of Geographic Information Studies, Vol. 13, No. 1, pp. 89-100. (in Korean with English abstract) https://doi.org/10.11108/KAGIS.2010.13.1.089
  11. Park, J.S., Lee, W.H., and Jo, M.H. (2016), Improving accuracy of land cover classification in river basins using Lansat-8 OLI image, vegetation index, and water index, Journal of Korean Association of Geographic Information Studies, Vol. 19, No. 2, pp. 98-106. (in Korean with English abstract) https://doi.org/10.11108/kagis.2016.19.2.098
  12. Zhu, J.Y., Park, T.S., Isola, P., and Efros., A.A. (2018), Unpaired image translation using cycle-consistent adversarial networks, arXiv:1703.10593.