DOI QR코드

DOI QR Code

Intelligent Face Recognition and Tracking System to Distribute GPU Resources using CUDA

쿠다를 사용하여 GPU 리소스를 분배하는 지능형 얼굴 인식 및 트래킹 시스템

  • Kim, Jae-Heong (Dept. Electronics&Control Engineering, Hanbat National University) ;
  • Lee, Seung-Ho (Dept. Electronics&Control Engineering, Hanbat National University)
  • Received : 2018.06.09
  • Accepted : 2018.06.19
  • Published : 2018.06.30

Abstract

In this paper, we propose an intelligent face recognition and tracking system that distributes GPU resources using CUDA. The proposed system consists of five steps such as GPU allocation algorithm that distributes GPU resources in optimal state, face area detection and face recognition using deep learning, real time face tracking, and PTZ camera control. The GPU allocation algorithm that distributes multi-GPU resources optimally distributes the GPU resources flexibly according to the activation level of the GPU, unlike the method of allocating the GPU to the thread fixedly. Thus, there is a feature that enables stable and efficient use of multiple GPUs. In order to evaluate the performance of the proposed system, we compared the proposed system with the non - distributed system. As a result, the system which did not allocate the resource showed unstable operation, but the proposed system showed stable resource utilization because it was operated stably. Thus, the utility of the proposed system has been demonstrated.

본 논문에서는 쿠다(CUDA)를 사용하여 GPU 리소스를 분배하는 지능형 얼굴 인식 및 트래킹 시스템을 제안한다. 제안한 시스템은 GPU 리소스를 최적의 상태로 분배하는 GPU 할당 알고리즘, 딥러닝을 이용한 얼굴 영역 검출, 딥러닝을 이용한 얼굴 인식, 실시간 얼굴 트래킹, PTZ 카메라 제어 등의 5단계로 구성되어진다. 멀티 GPU 리소스를 최적의 상태로 분배하는 GPU 할당 알고리즘은 고정적으로 스레드에 GPU를 할당하는 방식과 달리 GPU의 활성화 정도에 따라 유동적으로 GPU 리소스를 분배한다. 따라서 안정적이고 효율적인 멀티 GPU 사용을 가능하게 하는 특징이 있다. 제안된 시스템에 대한 성능을 평가하기 위하여 리소스 분배를 하지 않은 시스템과 제안한 시스템을 비교한 결과, 리소스를 분배하지 않은 시스템은 불안정한 동작을 보이는 반면에 제안한 시스템에서는 안정적으로 구동됨으로서 효율적인 리소스 사용을 보였다. 따라서 제안된 시스템의 효용성이 입증되었다.

Keywords

References

  1. Hee-Yeol Lee and Seung-Ho Lee, "A Study On Memory Optimization for Applying Deep Learning to PC," Journal of IKEEE, Vol. 21, No. 2, pp. 136-141, 2017.DOI:10.7471/ikeee.2017.21.2.136
  2. Lienhart, Rainer, and Jochen Maydt. "An extended set of haar-like features for rapid object detection." Image Processing. 2002. Proceedings. 2002 International Conference on. Vol. 1. IEEE, 2002.DOI: 10.1109/ICIP.2002.1038171
  3. Redmon, Joseph, et al. "You only look once: Unified, real-time object detection." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
  4. Redmon, Joseph. "Darknet: Open source neural networks in c." Pjreddie. com.[Online]. Available: https://pjreddie.com/darknet/.[Accessed: 21-Jun-2017] (2016).
  5. Jia, Yangqing, et al. "Caffe: Convolutional architecture for fast feature embedding." Proceedings of the 22nd ACM international conference on Multimedia. ACM, 2014. DOI:10.1145/2647868.2654889
  6. Lewis, John P. "Fast template matching." Vision interface. Vol. 95. No. 120123. 1995.
  7. Dalal, Navneet, and Bill Triggs. "Histograms of oriented gradients for human detection." Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on. Vol. 1. IEEE, 2005. DOI: 10.1109/CVPR.2005.177
  8. Hee-Yeol Lee, Sun-Gu Lee and Seung-Ho Lee, "Development of $360^{\circ}$ Omnidirectional IP Camera with High Resolution of 12Million Pixels," Journal of IKEEE, Vol. 21, No. 3, pp. 268-271, 2017 DOI:10.7471/ikeee.2017.21.3.26