| Peer-Reviewed

A Method of Infrared Image Pedestrian Detection with Improved YOLOv3 Algorithm

Received: 25 July 2021     Accepted: 9 August 2021     Published: 26 August 2021
Views:       Downloads:
Abstract

The principle of infrared image is thermal imaging technology. Infrared pedestrian detection technology can be applied to the safety monitoring of the elderly, which can not only protect personal privacy, but also realize pedestrian identification at night, which has strong application value and social significance. A method of infrared image pedestrian detection with improved YOLOv3 algorithm is proposed to increase the detection accuracy and solve the problem of low detection accuracy caused by infrared pedestrian target edge blurring. And according to the characteristics of infrared pedestrian, a complex sample data set is established which is applied to infrared pedestrian detection. The infrared image enhancement method with WDSR-B is adopted to improve the clarity of the data set. In addition, based on YOLOv3 algorithm, the output of the 4-time down-sampling layer is added to obtain richer context information for small targets and improve the detection performance of the network for small-target pedestrians. And the improved YOLOv3 network is trained by the enhanced infrared data set. Experimental results show that the scheme precision of pedestrian detection is higher than that of YOLOv3 algorithm. Therefore, this method can be applied to the detection of pedestrians at night and the safety monitoring of the elderly.

Published in American Journal of Optics and Photonics (Volume 9, Issue 3)
DOI 10.11648/j.ajop.20210903.11
Page(s) 32-38
Creative Commons

This is an Open Access article, distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium or format, provided the original work is properly cited.

Copyright

Copyright © The Author(s), 2021. Published by Science Publishing Group

Keywords

Infrared Image, Pedestrian Detection, Neural Network

References
[1] H. Krishna, and C. V. Jawahar, “Improving Small Object Detection,” 2017 4th IAPR Asian Conference on Pattern Recognition (ACPR), Nanjing, pp. 340-345 (2017).
[2] J. Shermeyer, and A. Van Etten, “The Effects of Super-Resolution on Object Detection Performance in Satellite Imagery,” 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Long Beach, CA, USA, pp. 1432-1441 (2019).
[3] C. Dong, C. C. Loy, K. He and X. Tang, “Image Super-Resolution Using Deep Convolutional Networks,” IEEE Transactions on Pattern Analysis and Machine Intelligence. 38 (2), 295-307, (2016).
[4] B. Lim, S. Son, H. Kim, S. Nah and K. M. Lee, “Enhanced Deep Residual Networks for Single Image Super-Resolution,” 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Honolulu, HI, pp. 1132-1140, (2017).
[5] Yu J, Fan Y, and Yang J, “Wide Activation for Efficient and Accurate Image Super-Resolution.” arXiv: 1808.08718v2 (2018).
[6] Z. Xie, S. Zhang, X. Yu, and G. Liu, “Infrared and visible face fusion recognition based on extended sparse representation classification and local binary patterns for the single sample problem,” J. Opt. Technol. 86 (13): 408-413, (2019).
[7] Axel-Christian Guei, and Moulay Akhloufi, “Deep learning enhancement of infrared face images using generative adversarial networks,” Appl. Opt. 57 (18), D98-D107, (2018).
[8] Wang C, and Qin S, “Approach for moving small target detection in infrared image sequence based on reinforcement learning.” Journal of Electronic Imaging, 25 (5): 053032, (2016).
[9] Redmon J, and Farhadi, “YOLOv3: An Incremental Improvement,” arXiv: 1804.02767, (2018).
[10] Xiangfu Zhang, Zhangsong Shi, Zhonghong Wu, and Jian Liu, “Sea surface ships detection method of UAV based on improved YOLOv3,” Proc. SPIE 11373, Eleventh International Conference on Graphics and Image Processing (ICGIP 2019), 113730T (2020).
[11] Tian, Wei, et al. “3D Pedestrian Detection in Farmland by Monocular RGB Image and Far-Infrared Sensing.” Remote Sensing [J] 13. 15 (2021): 2896.
[12] Zhang C, Li D, Qi J, et al. Infrared Small Target Detection Method with Trajectory Correction Fuze Based on Infrared Image Sensor [J]. Sensors, 2021, 21 (13): 4522.
[13] Shi W, Caballero J, and Ferenc Huszár, “Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network.” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1874-1883 (2016).
[14] Redmon J, Divvala S, and Girshick R, “You Only Look Once: Unified, Real-Time Object Detection,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 779-788 (2016).
[15] Redmon J, and Farhadi A, “YOLO9000: Better, Faster, Stronger,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE: 6517-6525 (2017).
[16] Guanglin Yang, “Output characteristics of pre-amplifying circuit signal for human body infrared detecting,” The 17th national academic annual meeting of measuring and controlling instruments (MCMI'2007), 87-90 (2007).
Cite This Article
  • APA Style

    Yue Sun, Yifeng Shao, Guanglin Yang, Haiyan Xie. (2021). A Method of Infrared Image Pedestrian Detection with Improved YOLOv3 Algorithm. American Journal of Optics and Photonics, 9(3), 32-38. https://doi.org/10.11648/j.ajop.20210903.11

    Copy | Download

    ACS Style

    Yue Sun; Yifeng Shao; Guanglin Yang; Haiyan Xie. A Method of Infrared Image Pedestrian Detection with Improved YOLOv3 Algorithm. Am. J. Opt. Photonics 2021, 9(3), 32-38. doi: 10.11648/j.ajop.20210903.11

    Copy | Download

    AMA Style

    Yue Sun, Yifeng Shao, Guanglin Yang, Haiyan Xie. A Method of Infrared Image Pedestrian Detection with Improved YOLOv3 Algorithm. Am J Opt Photonics. 2021;9(3):32-38. doi: 10.11648/j.ajop.20210903.11

    Copy | Download

  • @article{10.11648/j.ajop.20210903.11,
      author = {Yue Sun and Yifeng Shao and Guanglin Yang and Haiyan Xie},
      title = {A Method of Infrared Image Pedestrian Detection with Improved YOLOv3 Algorithm},
      journal = {American Journal of Optics and Photonics},
      volume = {9},
      number = {3},
      pages = {32-38},
      doi = {10.11648/j.ajop.20210903.11},
      url = {https://doi.org/10.11648/j.ajop.20210903.11},
      eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.ajop.20210903.11},
      abstract = {The principle of infrared image is thermal imaging technology. Infrared pedestrian detection technology can be applied to the safety monitoring of the elderly, which can not only protect personal privacy, but also realize pedestrian identification at night, which has strong application value and social significance. A method of infrared image pedestrian detection with improved YOLOv3 algorithm is proposed to increase the detection accuracy and solve the problem of low detection accuracy caused by infrared pedestrian target edge blurring. And according to the characteristics of infrared pedestrian, a complex sample data set is established which is applied to infrared pedestrian detection. The infrared image enhancement method with WDSR-B is adopted to improve the clarity of the data set. In addition, based on YOLOv3 algorithm, the output of the 4-time down-sampling layer is added to obtain richer context information for small targets and improve the detection performance of the network for small-target pedestrians. And the improved YOLOv3 network is trained by the enhanced infrared data set. Experimental results show that the scheme precision of pedestrian detection is higher than that of YOLOv3 algorithm. Therefore, this method can be applied to the detection of pedestrians at night and the safety monitoring of the elderly.},
     year = {2021}
    }
    

    Copy | Download

  • TY  - JOUR
    T1  - A Method of Infrared Image Pedestrian Detection with Improved YOLOv3 Algorithm
    AU  - Yue Sun
    AU  - Yifeng Shao
    AU  - Guanglin Yang
    AU  - Haiyan Xie
    Y1  - 2021/08/26
    PY  - 2021
    N1  - https://doi.org/10.11648/j.ajop.20210903.11
    DO  - 10.11648/j.ajop.20210903.11
    T2  - American Journal of Optics and Photonics
    JF  - American Journal of Optics and Photonics
    JO  - American Journal of Optics and Photonics
    SP  - 32
    EP  - 38
    PB  - Science Publishing Group
    SN  - 2330-8494
    UR  - https://doi.org/10.11648/j.ajop.20210903.11
    AB  - The principle of infrared image is thermal imaging technology. Infrared pedestrian detection technology can be applied to the safety monitoring of the elderly, which can not only protect personal privacy, but also realize pedestrian identification at night, which has strong application value and social significance. A method of infrared image pedestrian detection with improved YOLOv3 algorithm is proposed to increase the detection accuracy and solve the problem of low detection accuracy caused by infrared pedestrian target edge blurring. And according to the characteristics of infrared pedestrian, a complex sample data set is established which is applied to infrared pedestrian detection. The infrared image enhancement method with WDSR-B is adopted to improve the clarity of the data set. In addition, based on YOLOv3 algorithm, the output of the 4-time down-sampling layer is added to obtain richer context information for small targets and improve the detection performance of the network for small-target pedestrians. And the improved YOLOv3 network is trained by the enhanced infrared data set. Experimental results show that the scheme precision of pedestrian detection is higher than that of YOLOv3 algorithm. Therefore, this method can be applied to the detection of pedestrians at night and the safety monitoring of the elderly.
    VL  - 9
    IS  - 3
    ER  - 

    Copy | Download

Author Information
  • Laboratory of Signal and Information Processing, Department of Electronics, Peking University, Beijing, China

  • Optics Research Group, Department of Imaging Physics, Faculty of Applied Sciences, Delft University of Technology, Delft, Netherlands

  • Laboratory of Signal and Information Processing, Department of Electronics, Peking University, Beijing, China

  • Country China Science Patent and Trademark Agent, Beijing, China

  • Sections