DOI QR코드

DOI QR Code

카테고리 계층을 고려한 회선신경망의 이미지 분류

Image Classification Using Convolutional Neural Networks Considering Category Hierarchies

  • Jeong, Nokwon (IS Technology Inc.) ;
  • Cho, Soosun (Dept. of Computer Science and Information Eng., Korea National University of Transportation)
  • 투고 : 2018.08.16
  • 심사 : 2018.11.21
  • 발행 : 2018.12.31

초록

In order to improve the performance of image classifications using Convolutional Neural Networks (CNN), applying a category hierarchy to the classification can be a useful idea. However, the visual separation of object categories is very different according to the upper and lower category levels and highly uneven in image classifications. Therefore, it is doubtable whether the use of category hierarchies for classification is effective in CNN. In this paper, we have clarified whether the image classification using category hierarchies improves classification performance, and found at which level of hierarchy classification is more effective. For experiments we divided the image classification task according to the upper and lower category levels and assigned image data to each CNN model. We identified and compared the results of three classification models and analyzed them. Through the experiments, we could confirm that classification effectiveness was not improved by reduction of number of categories in a classification model. And we found that only with the re-training method in the last network layer, the performance of lower category classification was not improved although that of higher category classification was improved.

키워드

MTMDCW_2018_v21n12_1417_f0001.png 이미지

Fig. 1. Category hierarchy of upper & lower levels.

MTMDCW_2018_v21n12_1417_f0002.png 이미지

Fig. 2. Comparison of classification model A and B.

MTMDCW_2018_v21n12_1417_f0003.png 이미지

Fig. 3. Comparison of classification model B and C.

MTMDCW_2018_v21n12_1417_f0004.png 이미지

Fig. 4. Representative images occur classification errors.

Table 1. Experimental conditions

MTMDCW_2018_v21n12_1417_t0001.png 이미지

Table 2. Classification results from Validation sets

MTMDCW_2018_v21n12_1417_t0002.png 이미지

Table 3. Classification results from Test sets

MTMDCW_2018_v21n12_1417_t0003.png 이미지

Table 4. Comparison of True(T)/False(F) from model B and C

MTMDCW_2018_v21n12_1417_t0004.png 이미지

Table 5. Classification errors of model C.1

MTMDCW_2018_v21n12_1417_t0005.png 이미지

Table 6. Classification errors of model C affected by failures of model C.1

MTMDCW_2018_v21n12_1417_t0006.png 이미지

참고문헌

  1. A. Krizhevsky, I. Sutskever, and G.E. Hinton, "Imagenet Classification with Deep Convolution Neural Networks," Proceeding of Neural Information Processing System, pp. 1097-1105, 2012.
  2. K. Simonyan and A. Zisserman, "Very Deep Convolutional Networks for Large-Scale Image Recognition," Computer Vision and Pattern Recognition, arXiv:1409.1556, 2014.
  3. C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, "Rethinking the Inception Architecture for Computer Vision," Proceeding of 2016 IEEE Conference on Computer Vision and Pattern Recognition, arXiv:1512.00567, 2015.
  4. K. He, X. Zhang, S. Ren, and J. Sun, "Identity Mappings in Deep Residual Networks," Proceeding of European Conference on Computer Vision, arXiv:1603.05027, 2016.
  5. Y. Park, S. Gang, J. Chae, and J. Lee, "Classification Method of Plant Leaf Using Dense Net," Journal of Korea Multimedia Society, Vol. 21, No. 5, pp. 571-582, 2018. https://doi.org/10.9717/KMMS.2018.21.5.571
  6. H. Jang and S. Cho, "Automatic Tagging for Social Images Using Convolution Neural Networks," Journal of Korean Institute of Information Scientists and Engineers, Vol. 43, No. 1, pp. 47-53, 2016.
  7. N. Jeong and S. Cho, “Instagram Image Classification with Deep Learning,” Journal of Internet Computing and Services, Vol. 18, No. 5, pp. 61-67, 2017. https://doi.org/10.7472/jksii.2017.18.2.61
  8. A.M. Tousch, S. Herbin, and J.Y. Audibert, “Semantic Hierarchies for Image Annotation: A Survey,” Pattern Recognition, Vol. 45, No. 1, pp. 333-345, 2011. https://doi.org/10.1016/j.patcog.2011.05.017
  9. J. Deng, J. Krause, A.C. Berg, and L. Fei-Fei, "Hedging Your Bets: Optimizing Accuracy- Specificity Trade-Offs in Large Scale Visual Recognition," Proceeding of IEEE Conference on Computer Vision and Pattern Recognition, pp. 16-21, 2012.
  10. B. Liu, F. Sadeghi, M. Tappen, O. Shamir, and C. Liu, "Probabilistic Label Trees for Efficient Large Scale Image Classification," Proceeding of IEEE Conference on Computer Vision and Pattern Recognition, pp. 843-850, 2013.
  11. N. Srivastava and R. Salakhutdinov, "Discriminative Transfer Learning with Tree-Based Priors," Proceeding of Neural Information Processing System, pp. 2094-2102, 2013.
  12. J. Deng, N. Ding, Y. Jia, A. Frome, K. Murphy, S. Bengio et al., "Large-Scale Object Classification Using Label Relation Graphs," Proceeding of European Conference on Computer Vision, pp. 48-64, 2014.
  13. O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, et al., “ImageNet Large Scale Visual Recognition Challenge,” Journal of Computer Vision, Vol. 115, No. 3, pp. 211-252, 2015. https://doi.org/10.1007/s11263-015-0816-y