DOI QR코드

DOI QR Code

Object Detection using Fuzzy Adaboost

퍼지 Adaboost를 이용한 객체 검출

  • Received : 2016.03.18
  • Accepted : 2016.04.21
  • Published : 2016.05.28

Abstract

The Adaboost chooses a good set of features in rounds. On each round, it chooses the optimal feature and its threshold value by minimizing the weighted error of classification. The involved process of classification performs a hard decision. In this paper, we expand the process of classification to a soft fuzzy decision. We believe this expansion could allow some flexibility to the Adaboost algorithm as well as a good performance especially when the size of a training data set is not large enough. The typical Adaboost algorithm assigns a same weight to each training datum on the first round of a training process. We propose a new algorithm to assign different initial weights based on some statistical properties of involved features. In experimental results, we assess that the proposed method shows higher performance than the traditional one.

Adaboost 학습 알고리즘은 학습 단계마다 가장 좋은 특징을 선택하도록 하는 학습 알고리즘 이다. 각 학습 단계에서는 최적의 특징을 선택하기 위해 특정 임계값과 그에 대한 최소 오차율을 가지는 특징을 선택하도록 되어 있다. 하지만, 임계값을 이용하는 방법은 최적의 오차율을 검출하는데 있어 효율적인 방법이 아니다. 본 논문에서는 최적의 오차율을 검출하기 위한 퍼지 Adaboost 기법을 제안한다. 퍼지를 통해 결정 경계를 유연하게 한 Adaboost는 학습 단계가 적어도 좋은 성능을 보이는 장점이 있다. 기존의 Adaboost는 학습 전에 학습데이터에 대한 가중치를 동일하게 할당한다. 하지만, 본 논문에서는 이에 대한 가중치를 확률을 이용하여 초기 가중치를 다르게 줌으로서, 적은 학습에도 좋은 결과를 보이는 방법을 제안한다. 실험 결과에서는 기존의 Adaboost와 제안하는 방법에 대한 성능 평가를 통해, 퍼지 Adaboost가 기존 방법에 비해 좋은 결과를 보였다.

Keywords

References

  1. R. Rout, "A survey on object detection and tracking algorithms," PhD Thesis, National Institute of Technology Rourkela, 2013.
  2. D. Prasad, "Survey of the Problem of Object Detection in real images," International Journal of Image Processing, Vol.6, Issue.6, pp.441-466, 2012.
  3. S. Tong and D. Koller, "Support vector machine active learning with applications to text classification," The journal of machine learning research, pp.45-66, 2001.
  4. M. Hagan, H. Demuth, M. Beale, and O. Jesus, Neural Network Design, Boston:Pws Pub., 1996.
  5. P. Viola and M. Jones, "Rapid object detection using a boosted cascade of simple features," Computer Vision and Pattern Recognition, 2001.
  6. G. Ratsch, T. Onoda, and K. Muller, "Soft margins for AdaBoost," Machine learning, Vol.42, Issue.3, pp.287-320, 2001. https://doi.org/10.1023/A:1007618119488
  7. S. Joo, S. Weon, and H. Choi, "Real-time depth-based hand detection and tracking," The Scientific World Journal, 2014.
  8. R. Lienhart and M. Jochen, "An extened set of haar-like features for rapid object detection," Image Processing, 2002
  9. J. Zhu, S. Rosset, H. Zou and T. Hastie, "Multi-class adaboost," Statistics and its Inference, Vol.2, No.3, pp.349-360, 2009.
  10. W. Hu, J. Gao, Y. Wang, O. Wu, and S. Maybank, "Online adaboost-based parameterized methods for dynamic distributed network intrusion detection," Cybermetics, IEEE transactions on, Vol.44, No.1, pp.66-82, 2014. https://doi.org/10.1109/TCYB.2013.2247592
  11. G. Hinton, S. Osindero, and Y. Teh, "A fast learning algorithm for deep belief nets," Neural computation, Vol.18, No.7, pp.1527-1554, 2006. https://doi.org/10.1162/neco.2006.18.7.1527
  12. http://www.cs.nyu.edu/ roweis/data.html
  13. http://cbcl.mit.edu/software-datasets/Pedestrian Data.html
  14. S. Lomax and S. Vadera, "A survey of cost-sensitive decision tree induction algorithms," ACM Computing Surveys (CSUR), Vol.45, Issue.2, 2013.
  15. A. Graves, A. Mohamed, and G. Hinton, "Speech recognition with deep recurrent neural networks," Acoustics, Speech and Signal Processing (ICASSP), 2013.