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

Research and Optimization of Face Detection Algorithm Based on MTCNN Model in Complex Environment

복잡한 환경에서 MTCNN 모델 기반 얼굴 검출 알고리즘 개선 연구

  • Fu, Yumei (Department of Computer Engineering, Dong-Eui University) ;
  • Kim, Minyoung (Research Institute of ICT Fusion and Convergence, Dong-Eui University) ;
  • Jang, Jong-wook (Department of Computer Engineering, Dong-Eui University)
  • Received : 2019.09.30
  • Accepted : 2019.10.17
  • Published : 2020.01.31
Download PDF
⟨ Previous Next ⟩

Abstract

With the rapid development of deep neural network theory and application research, the effect of face detection has been improved. However, due to the complexity of deep neural network calculation and the high complexity of the detection environment, how to detect face quickly and accurately becomes the main problem. This paper is based on the relatively simple model of the MTCNN model, using FDDB (Face Detection Dataset and Benchmark Homepage), LFW (Field Label Face) and FaceScrub public datasets as training samples. At the same time of sorting out and introducing MTCNN(Multi-Task Cascaded Convolutional Neural Network) model, it explores how to improve training speed and Increase performance at the same time. In this paper, the dynamic image pyramid technology is used to replace the traditional image pyramid technology to segment samples, and OHEM (the online hard example mine) function in MTCNN model is deleted in training, so as to improve the training speed.

현재 심층 신경망 이론 및 응용 연구의 빠른 개발로 얼굴 인식의 효과가 향상되고 있다. 그러나 심층 신경망 계산의 복잡성과 탐지 환경의 복잡성으로 인해 얼굴을 빠르고 정확하게 감지하는 방법이 주요 문제가 된다. 이 논문은 FDDB, LFW 및 FaceScrub 공개 데이터 세트를 훈련 표본을 사용하는 단순한 MTCNN 모델을 기반으로 둔다. MTCNN 모델을 분류하고 소개하면서 학습 훈련 속도를 높이고 성능을 향상하는 방법을 모색합니다. 본 논문에서는 다이내믹 이미지 피라미드 기술을 이용하여 기존 이미지 Pyramid 기술을 대체하여 샘플을 분할하고 MTCNN 모델의 OHEM을 훈련에서 제거하여 훈련 속도를 향상시켰다.

Keywords

References

  1. H.Gan, C.Q, "Research on Face Recognition Algorithm Based on MT-CNN," Industrial control computer journal, vol. 31, no. 11, pp. 119-122, May. 2018.
  2. UMass Vision.Face Detection Data Set and Benchmark Home [Internet]. Available : http://vis-www.cs.umass.edu/fddb/.
  3. UMass Vision. Labeled Faces in the Wild [Internet]. Available : http://vis-www.cs.umass.edu/lfw/.
  4. vesion interaction group. FaceScrub [Internet]. Available : http://vintage.winklerbros.net/facescrub.html.
  5. J.Y.Wu, and S.Q.Chen, "An Face Detection Algorithm Base on Improved MTCNN," Software Guide journal, no. 09, pp. 23-26, Nov. 2019.
  6. W.Q.Zhao, H.Yan, and X.Q.Shao, "Object detection based on inproved non-maximum suppression algorithm," Journal of Image and Graphics, no. 11, pp. 1676-1685, Nov. 2018.
  7. D.Zhao, Q.C.Tang, and Z.B.Yu, "A Solution to Multi-Objective Optimization Problem with Improved Cross Entropy Optimization," Journal of XI'AN Jiao Tong University, vol. 53, no. 3, pp. 66-74, Mar. 2019.
  8. Z.Y.Tang, F.R.Meng, and Z.X.Wang, "Fast face recognition with regularized least square via sparse representation," Journal of Application Research of Computers, vol. 33, no. 9, pp. 2831-2836, Sep. 2016.
  9. X.J.Fan, S.B.Xuan, and T.Feng, "Behavior Recognition Based on Dropout Convolutional Neural Network," Journal of GuangXi University for nationalities, vol. 23, no. 1, pp. 76-78, Feb. 2017.