Skip to main content
Springer Nature Link
Log in

Fast–slow visual network for action recognition in videos

  • Published:
Multimedia Tools and Applications Aims and scope Submit manuscript

Abstract

The visual speed, as important visual information on videos, has the ability to enhance the performance of video action recognition. Modeling the visual speed of different actions facilitates their recognition. Previous works have often attempted to capture the visual speed through sampling raw videos at multiple rates and then constructing an input-level frame pyramid, which usually requires to manipulate a costly multibranched network. In this work, we proposed a fast–slow visual network (FSVN) to improve the accuracy of video action recognition via a visual speed stripping strategy which can flexibly be integrated into various excellent 2-D or 3-D backbone networks. Specifically, by the method of the frame difference, we divided a video into fast visual frames and slow visual frames which can respectively represent the action information and the spatial information in the video. Then, we designed a fast visual information recognition network to capture the action information and a slow visual information recognition network to record the spatial information; finally, these two networks were integrated. The experiments on the data sets UCF101 (98.3%) and HMDB51 (76.4%) prove the superiority of our method over the traditional approaches.

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

Access this article

Log in via an institution

Subscribe and save

Springer+ Basic
$34.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or eBook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

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

Similar content being viewed by others

References

  1. Avola D, Bernardi M, Foresti GL (2019) Fusing depth and colour information for human action recognition[J]. Multimed Tools Appl 78(5):5919–5939

    Article  Google Scholar 

  2. Carreira J, Zisserman A 2017 Quo vadis, action recognition? a new model and the kinetics dataset. in proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

  3. Chu H et al (2008) Target tracking algorithm based on camshift algorithm combined with difference in frame. Journal of Projectiles, Rockets, Missiles and Guidance 28(3):85–88

    Google Scholar 

  4. Deng J, Dong W, Socher R et al (2009) Imagenet: A large-scale hierarchical image database[C]. 2009 IEEE conference on computer vision and pattern recognition. Ieee 2009:248–255

    Google Scholar 

  5. Dhiman C, Vishwakarma DK (2020) View-invariant deep architecture for human action recognition using two-stream motion and shape temporal dynamics[J]. IEEE Trans Image Process 29:3835–3844

  6. Diba A, Sharma V, Van Gool L (2017) Deep temporal linear encoding networks[C]. Proceedings of the IEEE conference on Computer Vision and Pattern Recognition 2017:2329–2338

  7. Feichtenhofer C, Pinz A, Zisserman A (2016) Convolutional two-stream network fusion for video action recognition[C]. Proceedings of the IEEE conference on computer vision and pattern recognition 2016:1933–1941

  8. Feichtenhofer C, Fan H, Malik J et al (2019) Slowfast networks for video recognition[C]. Proceedings of the IEEE/CVF international conference on computer vision 2019:6202–6211

  9. Ge H, Yan Z, Yu W et al (2019) An attention mechanism based convolutional LSTM network for video action recognition[J]. Multimed Tools Appl 78(14):20533–20556

    Article  Google Scholar 

  10. He K, Zhang X, Ren S et al (2016) Deep residual learning for image recognition[C]. Proceedings of the IEEE conference on computer vision and pattern recognition 2016:770–778

  11. Kuehne H, Jhuang H, Garrote E et al (2011) HMDB: a large video database for human motion recognition[C]. 2011 International conference on computer vision. IEEE 2011:2556–2563

  12. Kashiwagi T, Oe S, Terada K (2000) Edge characteristic of color image and edge detection using color histogram. IEEJ Transactions on Electronics, Information and Systems 120(5):715–723

    Article  Google Scholar 

  13. Kay W, Carreira J, Simonyan K et al (2017) The kinetics human action video dataset[J]. arXiv preprint arXiv:1705.06950

  14. Kumar K (2019) EVS-DK: Event video skimming using deep keyframe[J]. J Vis Commun Image Represent 58:345–352

  15. Kumar K, Shrimankar DD (2018) ESUMM: event summarization on scale-free networks[J]. IETE Technical Review

  16. Kumar K, Shrimankar DD, Singh N (2018) V-LESS: a video from linear event summaries[C]. Proceedings of 2nd International Conference on Computer Vision & Image Processing. Springer, Singapore, pp 385–395

  17. Kumar K, Shrimankar DD, Singh N (2019) Key-lectures: keyframes extraction in video lectures[M]//Machine intelligence and signal analysis. Springer, Singapore, pp 453–459

  18. Lan Z, Zhu Y, Hauptmann AG et al (2017) Deep local video feature for action recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition workshops 2017:1–7

  19. Peng L, Lafortune EPF, Greenberg DP et al (1997) Use of computer graphic simulation to explain color histogram structure[C]. Color and Imaging Conference. Society for Imaging Science and Technology 1997(1):187–192

  20. Pengcheng D, Siyuan C, Zhenyu Z et al (2019) Human Behavior Recognition Based on IC3D[C]. 2019 Chinese Control And Decision Conference (CCDC). IEEE 2019:3333–3337

  21. Qiu Z, Yao T, Mei T (2017) Learning spatio-temporal representation with pseudo-3d residual networks[C]. Proceedings of the IEEE International Conference on Computer Vision 2017:5533–5541

  22. Simonyan K, Zisserman A (2014) Two-stream convolutional networks for action recognition in videos[J]. Advances in neural information processing systems, 27

  23. Simonyan K, Zisserman A, 2014 Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556

  24. Solanki A, Bamrara R, Kumar K et al (2020) VEDL: a novel video event searching technique using deep learning[M]. Soft Computing: Theories and Applications. Springer, Singapore, pp 905–914

  25. Soomro K, Zamir AR, Shah M 2012 A dataset of 101 human action classes from videos in the wild. Center for Research in Computer Vision, 2(11)

  26. Sun L, Jia K, Yeung DY et al (2015) Human action recognition using factorized spatio-temporal convolutional networks[C]. Proceedings of the IEEE international conference on computer vision 2015:4597–4605

  27. Szegedy C, et al. 2015 Going deeper with convolutions. in Proceedings of the IEEE conference on computer vision and pattern recognition

  28. Tang Q, Dai S.G, Yang J 2013 Object tracking algorithm based on camshift combining background subtraction with three frame difference. In applied mechanics and materials. 2013. Trans Tech Publ

  29. Tran D, Bourdev L, Fergus R et al (2015) Learning spatiotemporal features with 3d convolutional networks[C]. Proceedings of the IEEE international conference on computer vision 2015:4489–4497

  30. Wang H, Schmid C (2013) Action recognition with improved trajectories[C]. Proceedings of the IEEE international conference on computer vision 2013:3551–3558

  31. Wang L, Qiao Y, Tang X (2015) Action recognition with trajectory-pooled deep-convolutional descriptors[C]. Proceedings of the IEEE conference on computer vision and pattern recognition 2015:4305–4314

  32. Wang L, Xiong Y, Wang Z et al (2016) Temporal segment networks: Towards good practices for deep action recognition[C]. European conference on computer vision. Springer, Cham, pp 20–36

  33. Wu H, Liu J, Zhu X et al (2021) Multi-scale spatial-temporal integration convolutional tube for human action recognition[C]. Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence 2021:753–759

  34. Xu Y, Chen M, Xie T (2017) Method for state recognition of egg embryo in vaccines production based on support vector machine[J]. DEStech Transactions on Engineering and Technology Research, (tmcm)

  35. Yang C, Xu Y, Shi J et al (2020) Temporal pyramid network for action recognition[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020:591–600

  36. Yoo GH, Park JM, You KS et al (2005) Content-Based Image Retrieval Using Adaptive Color Histogram[J]. The Journal of Korean Institute of Communications and Information Sciences 30(9C):949–954

    Google Scholar 

  37. Zhang D, Dai X, Wang YF (2018) Dynamic temporal pyramid network: a closer look at multi-scale modeling for activity detection[C]. Asian Conference on Computer Vision. Springer, Cham, pp 712–728

  38. Zhong X, Tu K, Xia H (2017) Mean-shift algorithm fusing multi feature[C]. 2017 IEEE 2nd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC). IEEE 2017: 245–1249

Download references

Acknowledgements

This research was jointly supported by (1) National Natural Science Foundation of China (U1809208); (2) the Research and Development Fund Talent Startup Project of Zhejiang A&F University (No. 2019FR070).

Author information

Authors and Affiliations

  1. School of Information Engineering, Zhejiang A & F University, Hangzhou, China

    Heng Hu, Tongcun Liu & Hailin Feng

  2. Key Laboratory of Forestry Intelligent Monitoring and Information Technology of Zhejiang Province, Hangzhou, China

    Heng Hu, Tongcun Liu & Hailin Feng

  3. Key Laboratory of State Forestry Administration on Forestry Sensing Technology and Intelligent Equipment, Hangzhou, China

    Heng Hu, Tongcun Liu & Hailin Feng

Authors
  1. Heng Hu
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Tongcun Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Hailin Feng
    View author publications

    You can also search for this author inPubMed Google Scholar

Corresponding author

Correspondence to Hailin Feng.

Ethics declarations

Conflict of interest

We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.

Additional information

Publisher’s note

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

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hu, H., Liu, T. & Feng, H. Fast–slow visual network for action recognition in videos. Multimed Tools Appl 81, 26361–26379 (2022). https://doi.org/10.1007/s11042-022-12948-3

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11042-022-12948-3

Keywords