Moving Objects Segmentation Based on DeepSphere in Video Surveillance

  • Sirine AmmarEmail author
  • Thierry Bouwmans
  • Nizar Zaghden
  • Mahmoud Neji
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 11845)


Segmentation of moving objects from video sequences plays an important role in many computer vision applications. In this paper, we present a background subtraction approach based on deep neural networks. More specifically, we propose to employ and validate an unsupervised anomaly discovery framework called “DeepSphere” to perform foreground objects detection and segmentation in video sequences. DeepSphere is based on both deep autoencoders and hypersphere learning methods to isolate anomaly pollution and reconstruct normal behaviors in spatial and temporal context. We exploit the power of this framework and adjust it to perform foreground objects segmentation. We evaluate the performance of our proposed method on 9 surveillance videos from the Background Model Challenge (BMC 2012) dataset, and compare that with a standard subspace learning technique, Robust Principle Component Analysis (RPCA) as well as a Deep Probabilistic Background Model (DeepPBM). Experimental results show that our approach achieved successful results than other existing ones.


Background subtraction Unsupervised anomaly discovery DeepSphere DeepPBM RPCA 


  1. 1.
    Shuigen W., Zhen C., and Hua D. Motion detection based on temporal difference method and optical flow field. In: Second International Symposium on Electronic Commerce and Security, pp. 85–88 (2009)Google Scholar
  2. 2.
    Mahraz, M.A., Riffi, J., Tairi, H.: Motion estimation using the fast and adaptive bidimensional empirical mode decomposition. J. Real-Time Image Process. 9(3), 491–501 (2014)CrossRefGoogle Scholar
  3. 3.
    Candès, E., Li, X., Ma, Y., Wright, J.: Robust principal component analysis? Int. J. ACM 58(3), 11 (2011)MathSciNetzbMATHGoogle Scholar
  4. 4.
    Wren, C., Azarbayejani, A.: Pfinder: real-time tracking of the human body. IEEE Trans. Pattern Anal. Mach. Intell. 19, 780–785 (1997)CrossRefGoogle Scholar
  5. 5.
    Pulgarin-Giraldo, J., Alvarez-Meza, A., Insuasti-Ceballos, D., Bouwmans, T., Castellanos-Dominguez, G.: GMM background modeling using divergence-based weight updating. In: Conference Ibero-American Congress on Pattern Recognition (2016)Google Scholar
  6. 6.
    Zivkovic, Z.: Efficient adaptive density estimation per image pixel for the task of background subtraction. Pattern Recogn. Lett. 27(7), 773–780 (2006)CrossRefGoogle Scholar
  7. 7.
    Bouwmans, T., El Baf, F.: Modeling of dynamic backgrounds by type-2 fuzzy gaussians mixture models. J. Basic Appl. Sci. 1(2), 265–276 (2009)Google Scholar
  8. 8.
    Zhang, H., Xu D.: Fusing color and gradient features for background model. In: International Conference on Signal Processing, vol. 2, no. 7 (2006)Google Scholar
  9. 9.
    Baf, F.E., Bouwmans, T., Vachon, B.: Fuzzy integral for moving object detection. In: IEEE International Conference on Fuzzy Systems, pp. 1729–1736 (2008)Google Scholar
  10. 10.
    Oliver, N., Rosario, B., Pentland, A.: A Bayesian computer vision system for modeling human interactions. In: ICVS (1999)Google Scholar
  11. 11.
    Dong, Y., DeSouza, G.: Adaptive learning of multi-subspace for foreground detection under illumination changes. Comput. Vis. Image 115, 31–49 (2011) CrossRefGoogle Scholar
  12. 12.
    Bouwmans, T., Sobral, A., Javed, S., Jung, S.: Decomposition into low-rank plus additive matrices for background/foreground separation:a review for a comparative evaluation with a large-scale dataset. Comput. Sci. Rev. 23, 1–71 (2017)CrossRefGoogle Scholar
  13. 13.
    Guyon, C., Bouwmans, T., Zahzah, E.: Foreground detection by robust PCA solved via a linearized alternating direction method. In: International Conference on Image Analysis and Recognition, ICIAR (2012)Google Scholar
  14. 14.
    Vaswani, N., Bouwmans, T., Javed, S., Narayanamurth, P.: Robust subspace learning: robust PCA, robust subspace tracking and robust subspace recovery. IEEE Signal Process. Mag. 35(4), 32–55 (2018b)CrossRefGoogle Scholar
  15. 15.
    Xu, P., Ye, M., Liu, Q., Li, X., Ding, J.: Motion detection via a couple of auto-encoder networks. In: International Conference on Multimedia and Expo, ICME (2014)Google Scholar
  16. 16.
    Braham, M., Droogenbroeck, M.V.: Deep background subtraction with scene-specific convolutional neural networks. In: International Conference on Systems, Signals and Image Processing, IWSSIP, pp. 1–4 (2016)Google Scholar
  17. 17.
    Wang, Y., Luo, Z., Jodoin, P.: Interactive deep learning method for segmenting moving objects. Pattern Recogn. Lett. 96, 66–75 (2016)CrossRefGoogle Scholar
  18. 18.
    Babaee, M., Dinh, D., Rigoll, G.: A deep convolutional neural network for background subtraction. Pattern Recogn. (2017) Google Scholar
  19. 19.
    Lim, K., Jang, W., Kim, C.: Background subtraction using encoder-decoder structured convolutional neural network. In: IEEE International Conference on Advanced Video and Signal Based Surveillance, AVSS (2017)Google Scholar
  20. 20.
    Zhang, Y., Li, X., Zhang, Z., Wu, F., Zhao, L.: Deep learning driven blockwise moving object detection with binaryscene modeling. Neurocomputing 168, 454–463 (2015)CrossRefGoogle Scholar
  21. 21.
    Zhou, C., Paffenroth, R.: Anomaly detection with robust deep autoencoders. In: KDD (2017)Google Scholar
  22. 22.
    Chalapathy, R., Menon, A., Chawla, S.: Robust, deep and inductive anomaly detection. Preprint (2017)Google Scholar
  23. 23.
    Teng, X., Yan, M., Ertugrul, A., Lin, Y.: Deep into hypersphere: robust and unsupervised anomaly discovery in dynamic networks. In: International Joint Conference on Artificial Intelligence, IJCAI, pp. 2724–2730 (2018)Google Scholar
  24. 24.
    Cun, Y.L., Bottou, L., Haffner, P.: Gradient-based learning applied to document recognition. Proc. IEEE 86, 2278–2324 (1998)CrossRefGoogle Scholar
  25. 25.
    Bautista, C.M., Dy, C.A., Manalac, M.I., Orbe, R.A., Cordel, M.: Convolutional neural network for vehicle detection in low resolution traffic videos. TENSYMP 9(11), 277–281 (2016)Google Scholar
  26. 26.
    Yan, Y., Zhao, H., Kao, F., Vargas, V., Zhao, S., Ren, J.: Deep background subtraction of thermal and visible imagery for pedestrian detection in videos. In: International Conference on Brain Inspired Cognitive Systems, BICS (2018)Google Scholar
  27. 27.
    Lim, L., Keles, H.: Foreground segmentation using a triplet convolutional neural network for multiscale feature encoding. Preprint (2018)Google Scholar
  28. 28.
    Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. In: International Conference on Learning Representation (2015)Google Scholar
  29. 29.
    Lim, L., Keles, H.: Foreground segmentation using convolutional neural networks for multiscale feature encoding. Pattern Recogn. Lett. 112, 256–262 (2018)CrossRefGoogle Scholar
  30. 30.
    Lim, L., Ang, l., Keles, H.: Learning multi-scale features for foreground segmentation. Preprint (2018)Google Scholar
  31. 31.
    Liao, J., Guo, G., Yan, Y., Wang, H.: Multiscale cascaded scene-specific convolutional neural networks for background subtraction. In: Pacific Rim Conference on Multimedia, PCM, pp. 524–533 (2018)Google Scholar
  32. 32.
    Bakkay, M., Rashwan, H., Salmane, H., Khoudour, L., Puig, D., Ruichek, Y.: BSCGAN: deep background subtraction with conditional generative adversarial networks. In: International Conference on Image Processing, ICIP (2018)Google Scholar
  33. 33.
    Zheng, W., Wang, K., Wang, F.: Background subtraction algorithm based on Bayesian generative adversarial networks. Acta Automatica Sinica 44, 878–890 (2018)zbMATHGoogle Scholar
  34. 34.
    Bahri, F., Shakeri, M., Ray, N.: Online illumination invariant moving object detection by generative neural network. Preprint (2018)Google Scholar
  35. 35.
    Choo, S., Seo, W., Jeong, D., Cho, N.: Multi-scale recurrent encoder-decoder network for dense temporal classification. In: ICPR, pp. 103–108 (2018)Google Scholar
  36. 36.
    Farnoosh, A., Rezaei, B., Ostadabbas, S.: DeepPBM: deep probabilistic background model estimation from video sequences. Preprint (2019)Google Scholar
  37. 37.
    Doersch, C.: Tutorial on variational autoencoders. Preprint (2016)Google Scholar
  38. 38.
    Dai, J., Wang, Y., Aston, J., Wipf, D.: Connections with robust PCA and the role of emergent sparsity in variational autoencoder models. J. Mach. Learn. Res. (JMLR) 19, 1–42 (2018)MathSciNetzbMATHGoogle Scholar
  39. 39.
    Liu, X., Zhao, G., Yao, J., Qi, C.: Background subtraction based on low-rank and structured sparse decomposition. Trans. Image Process 24, 2502–2514 (2015)MathSciNetCrossRefGoogle Scholar
  40. 40.
    Ammar, S., Zaghden, N., Neji, M.: A framework for people re-identification in multi-camera surveillance systems. In: International Conference on Cognition and Exploratory Learning in Digital Age CELDA (2017)Google Scholar
  41. 41.
    Vacavant, A., Chateau, T., Wilhelm, A., Lequièvre, L.: A benchmark dataset for outdoor foreground/background extraction. In: Park, J.-I., Kim, J. (eds.) ACCV 2012. LNCS, vol. 7728, pp. 291–300. Springer, Heidelberg (2013). Scholar
  42. 42.
    Baytas, M., Xiao, C., et al.: Patient subtyping via time-aware LSTM networks. In: SIGKDD, vol. 19, pp. 65–74 (2017)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  • Sirine Ammar
    • 1
    • 2
    Email author
  • Thierry Bouwmans
    • 2
  • Nizar Zaghden
    • 1
  • Mahmoud Neji
    • 1
  1. 1.Lab. MIRACLUniv de SfaxSfaxTunisia
  2. 2.Lab. MIAUniv de La RochelleLa RochelleFrance

Personalised recommendations