Skip to main content
SpringerLink
Log in

Fast and efficient face detector based on large kernel attention for CPU device

  • Research
  • Published:
Journal of Real-Time Image Processing Aims and scope Submit manuscript

Abstract

Efficient face detectors with low computational cost and fast speed are still a pressing problem despite the significant progress made in uncontrolled face detection. To address this issue, we propose two lightweight face detectors for CPU device, named speed-priority face detector (SPFD) and accuracy-priority face detector (APFD). In the case of SPFD, we propose a simplified version of FaceBoxes by reducing convolutional layers and channel number for reducing model complexity and computational cost, as well as replacing max pooling layers with group convolutional layers to learn local information and features’ integration. Additionally, large kernel attention modules based on prior knowledge of face size and network architecture are applied to increase the expression ability of features and capture the key information from long distances. Finally, an iterative retrained method is designed to further enhance the accuracy without increasing any cost during model testing. Regarding APFD, a new anchor generation strategy is utilized to find out more faces based on SPFD. Extensive experiments conducted on WIDER FACE validation dataset indicate that our detectors exceed FaceBoxes comprehensively in terms of accuracy and speed. Specifically, SPFD outperforms FaceBoxes by 4.1%, 8.6%, and 9.6% on WIDEFACE validation dataset while achieving the fastest detection speed on CPU device for VGA-resolution images. The average speed can reach 45FPS on CPU devices, while the size of its parameters is only 0.66 times of FaceBoxes. Moreover, the APFD outperforms many famous and lightweight face detectors and attains superior accuracy (easy: 91.4%, medium:88.1%, and hard: 64.7%) with 20FPS; it achieves the best trade-off between accuracy and speed for face detection.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13

Similar content being viewed by others

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  1. Zhang, K., Zhang, Z., Li, Z., Qiao, Y.: Joint face detection and alignment using multitask cascaded convolutional networks. IEEE Signal Process. Lett. 23(10), 1499–1503 (2016). https://doi.org/10.1109/LSP.2016.2603342

    Article  Google Scholar 

  2. Zhang, S., Zhu, X., Lei, Z., Shi, H., Wang, X., Li, S.Z.: FaceBoxes: a CPU real-time face detector with high accuracy. In: 2017 IEEE International Joint Conference on Biometrics, Denver, CO, USA, pp. 1–9. 10.1109/BTAS.2017.8272675 (2017)

  3. He, Y., Xu, D., Wu, L. et al. LFFD: A Light and Fast Face Detector for Edge Devices . arXiv1904.10633 (2019)

  4. Yoon, Jongmin, Kim, Daijin: An accurate and real-time multi-view face detector using ORFs and doubly domain-partitioning classifier. J. Real-Time Image Process. 16(6), 2425–2440 (2019)

    Article  Google Scholar 

  5. Yang, Z., Li, J., Min, W., et al.: Real-time pre-identification and cascaded detection for tiny faces. Appl. Sci. 9(20), 4344 (2019)

    Article  Google Scholar 

  6. Zhang, H., Wang, X., Zhu, J., JayKuo, C.-C.: Fast face detection on mobile devices by leveraging global and local facial characteristics. Signal Process. Image Commun. 78, 1–8 (2019). https://doi.org/10.1016/j.image.2019.05.016

    Article  Google Scholar 

  7. RetinaFaceJ. Deng, J., Guo, Y., Zhou, J., Yu, I.K., Zafeiriou, S.: RetinaFace: Single-Stage Dense Face Localisation in the Wild. arXiv1905.00641 (2019)

  8. Howard, A.G., Zhu, M., Chen, B., Kalenichenko, D.: MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications. arXiv1704.04861 (2017)

  9. Guo, J., Deng, J., Liu, Z., Lattas, A., Zafeiriou, S.: Sample and Computation Redistribution for Efficient Face Detection. arXiv:2105.04714 (2021)

  10. Liu, Z., Deng, J., Wang, F ., Shang, L., Xie, X., Sun, B.: DamoFD: digging into backbone design on face detection. In: 2023 IEEE/CVF Conference on International Conference on Learning Representations, Kigali, Rwanda, KGL (2023)

  11. Qi, D., Tan, W., Yao, Q. , Liu, J.: YOLO5Face: why reinventing a face detector. In: Lecture Notes in Computer Science, Vol. 13805 LNCS, pp. 228–244. arXiv:2105.12931 (2021)

  12. Xu, Y. et al. DBface. https://github.com/dlunion/dbface (2020)

  13. Xu, Y., Yan, W., Yang, G., et al.: CenterFace: joint face detection and alignment using face as point. Sci. Program. (2019). https://doi.org/10.1016/j.image.2019.05.016

    Article  Google Scholar 

  14. Linzai, et al. Ultraface. https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB (2020)

  15. Feng, Y., Yu, S., Peng, H., Li, Y.-R., Zhang, J.: Detect faces efficiently: a survey and evaluations. IEEE Trans. Biometrics Behav. Identity Sci. 4(1), 1–18 (2022). https://doi.org/10.1109/TBIOM.2021.3120412

    Article  Google Scholar 

  16. hpc203: https://github.com/hpc203/10kinds-light-face-detector-align-recognition (2021)

  17. Guo, M., Lu, C., Liu, Z., Cheng, M., Hu, S.: Visual Attention Network. arXiv:2202.09741 (2022)

  18. Yang, S., Luo, P., Loy, C. C., Tang, X.: WIDER FACE: a face detection benchmark. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, pp. 5525–5533. 10.1109/CVPR.2016.596 (2016)

  19. Jain, Vidit, Learned-Miller, Erik: Technical Report UM-CS-2010-009. FDDB: a benchmark for face detection in unconstrained settings, Dept. of Computer Science, University of Massachusetts, Amherst (2010)

    Google Scholar 

  20. Zhu, X., Ramanan, D.: Face detection, pose estimation, and landmark localization in the wild. In: 2012 IEEE Conference on Computer Vision and Pattern Recognition, June 16–21, 2012, pp. 2879–2886 (2012)

  21. Yan, J., Zhang, X., Lei, Z., Li, S.Z.: Face detection by structural models. In: 2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition, vol. 32(10), pp. 790–799 (2014)

  22. Qiao, S., Chen, L.-C., Yuille, A.: DetectoRS: detecting objects with recursive feature pyramid and switchable Atrous convolution. In: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, pp. 10208–10219. https://doi.org/10.1109/CVPR46437.2021.01008 (2021)

  23. Yu, J., Jiang, Y., Wang, Z., Cao, Z., Huang, T.: UnitBox: an advanced object detection network. In: Proceedings of the 2016 ACM multimedia conference, New York, NY, USA, pp. 516–520. 10.1145/2964284.2967274 (2016)

  24. Dong, C., Ren, S., Wei, Y., Cao, X., Jian, S.: Joint cascade face detection and alignment. In: 2014 European Conference on Computer Vision, Zurich, Switzerland, pp. 109–122. 10.1007/978-3-319-10599-4_8 (2014)

  25. Yang, S., Luo, P., Loy, C. , Tang, X.: From facial parts responses to face detection: a deep learning approach. In: 2015 IEEE International Conference on Computer Vision, Santiago, Chile, pp. 3676–3684. 10.1109/ICCV.2015.419 (2015)

  26. Ohn-Bar, E., Trivedi, M.M.: To boost or not to boost? On the limits of boosted trees for object detection. In: 2016 23rd International Conference on Pattern Recognition, Cancun, Mexico, pp. 3350–3355. 10.1109/ICPR.2016.7900151 (2016)

  27. Triantafyllidou, Danai, Tefas, Anastasios: A fast deep convolutional neural network for face detection in big visual data. Adv. Intell. Syst. Comput. 529, 61–70 (2016). https://doi.org/10.1007/978-3-319-47898-2_7

    Article  Google Scholar 

  28. https://github.com/ShiqiYu/libfacedetection.train (2021)

  29. Mathias, M., Benenson, R., Pedersoli, M., Gool, L.J.V.: Face detection without bells and whistles. Eur. Conf. Comput. Vis. 8692, 720–735 (2014)

    Google Scholar 

  30. Yan, J., Zhang, X., Lei, Z., Yi, D., Li, S.Z.: Structural models for face detection. In: 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG), Shanghai, China, vol. 2013, pp. 1–6 (2013). https://doi.org/10.1109/FG.2013.6553703

Download references

Acknowledgements

This work was supported in part by the National Natural Science Foundation of China (Grant Nos 61872448) and the Natural Science Basic Research Plan in Shanxi Province of China (No. 2021JQ-379).

Author information

Authors and Affiliations

  1. Educational Administration Department, National University of Defense Technology, Changsha, 410073, China

    Shuaihui Qi

  2. Test Center, National University of Defense Technology, Xian, 710106, China

    Xiaofeng Song

  3. Geely University of China, Chengdu, 641423, China

    Zhiyuan Li

  4. College of Information and Communication, National University of Defense Technology, Wuhan, 430000, China

    Tao Xie

Authors
  1. Shuaihui Qi
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Xiaofeng Song
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zhiyuan Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Tao Xie
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

SQ and XS proposed the network structure. SQ and ZL wrote the program code and undertake experimental work. All the authors wrote the main manuscript text and reviewed the manuscript.

Corresponding authors

Correspondence to Xiaofeng Song or Zhiyuan Li.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Qi, S., Song, X., Li, Z. et al. Fast and efficient face detector based on large kernel attention for CPU device. J Real-Time Image Proc 20, 72 (2023). https://doi.org/10.1007/s11554-023-01326-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s11554-023-01326-3

Keywords

Search

Navigation