access icon free Orientation truncated centre learning for deep face recognition

© The Institution of Engineering and Technology
Received 17/04/2018, Published 10/08/2018

Recently, centre loss that aiming to assist Softmax loss with the objectives of both inter-class dispension and intra-class compactness simultaneously, has achieved remarkable performance on convolutional neural network-based face recognition. However, its advantages highly rely on the centre feature assumption, which influences the capacity of the final obtained face features. Inspired by the centre loss approach, a novel Orientation Truncated Centre Learning is proposed, which takes advantage of an orientation truncated centre function to make the centre feature learning have more suitable orientation for deep face recognition. Three metrics are proposed to evaluate how discriminative are the distributions of the learned features for MNIST visualisation. Experimental results on several challenging benchmarks, including fine-grained labelled faces in the wild (FGLFW), labelled faces in the wild (LFW), YouTube faces (YTF), and benchmark of large-scale unconstrained face recognition (BLUFR), show that the proposed approach can easily generate more favourable results than several state-of-the-art competitors.

Inspec keywords: face recognition; learning (artificial intelligence); feedforward neural nets; feature extraction

Other keywords: orientation truncated centre function; centre feature assumption; centre loss; centre feature learning; orientation truncated centre learning; interclass dispension; MNIST visualisation; convolutional neural network-based face recognition; LFW benchmark; BLUFR benchmark; FGLFW benchmark; YTF benchmark; intraclass compactness; Softmax loss; deep face recognition

Subjects: Neural computing techniques; Knowledge engineering techniques; Image recognition; Computer vision and image processing techniques

References

    1. 1)
      • 5. Chen, B.H., Deng, W.H., Du, J.P.: ‘Noisy softmax: improving the generalization ability of dcnn via postponing the early softmax saturation’. Proc. of IEEE CVPR, Honolulu, HI, USA, July 2017.
    2. 2)
    3. 3)
      • 3. Wen, Y.D., Zhang, K.P., Li, Z.F., et al: ‘A discriminative feature learning approach for deep face recognition’. Proc. of ECCV, Amsterdam, The Netherlands, October 2016, pp. 499–515.
    4. 4)
      • 7. Parkhi, O., Vedaldi, A., Zisserman, A., et al: ‘Deep face recognition’, Proc. BMVC, 2015, 1, p. 6.
    5. 5)
      • 9. Liu, W.Y., Wen, Y.D., Yu, Z.D., et al: ‘Sphereface: deep hypersphere embedding for face recognition’. Proc. of IEEE CVPR, Honolulu, HI, USA, July 2017, pp. 212–220.
    6. 6)
      • 4. Yi, D., Lei, Z., Liao, S.C., et al: ‘Learning face representation from scratch’. arXiv preprint arXiv:1411.7923, 2014.
    7. 7)
      • 10. Wang, F., Xiang, X., Cheng, J., et al: ‘Normface: L2 hypersphere embedding for face verification’. Proc. of ACM Multimedia, Mountainview, CA, USA, October 2017.
    8. 8)
      • 2. Schroff, F., Kalenichenko, D., Philbin, J.: ‘Facenet: a unified embedding for face recognition and clustering’. Proc. of IEEE CVPR, Boston, MA, June 2015, pp. 815–823.
    9. 9)
      • 11. Wu, X., He, R., Sun, Z.N.: ‘A lightened cnn for deep face representation’, arXiv preprint, arXiv:1511.02683, 2015.
    10. 10)
      • 1. Sun, Y., Chen, Y.H., Wang, X.G., et al: ‘Deep learning face representation by joint identification-verification’. Proc. of NIPS, Montreal, QC, Canada, December 2014, pp. 1988–1996.
    11. 11)
      • 8. Taigman, Y., Yang, M., Ranzato, M., et al: ‘Deepface: closing the gap to human-level performance in face verification’. Proc. of IEEE CVPR, Columbus, OH, USA, June 2014, pp. 1701–1708.
http://iet.metastore.ingenta.com/content/journals/10.1049/el.2018.1326
Loading

Related content

content/journals/10.1049/el.2018.1326
pub_keyword,iet_inspecKeyword,pub_concept
6
6
Loading