access icon free Mask-guided class activation mapping network for person re-identification

© The Institution of Engineering and Technology
Received 30/05/2020, Published 04/11/2020

In this Letter, the authors propose a novel mask-guided class activation mapping (MCAM) network for person re-identification, which learns background-invariant and view-invariant features. Specifically, a novel loss function named mask-guided mapping loss is meticulously formulated to utilise the human binary masks, which contain helpful body shape information as the reference standard, thereby guiding the model to place more emphasis on human body regions. Moreover, they propose a new weighted channel attention (WCA) module, which replaces the global average pooling with a global depthwise convolution layer. By virtue of this particular WCA module, the feature information distributed across the spatial space can be individually weighted and dynamically compressed into a more precise channel attention map. Extensive experiments have been carried out on three widely-used re-identification data sets. Compared with the baseline model, MCAM has gained rank-1 accuracy improvement of 2.0% on Market-1501, 6.0% on DukeMTMC-reID, and 7.5% on CUHK03-NP, confirming its effectiveness.

Inspec keywords: image recognition; feature extraction; learning (artificial intelligence)

Other keywords: re-identification data sets; background-invariant features; view-invariant features; human binary masks; channel attention map; global average pooling; body shape information; WCA module; weighted channel attention module; human body regions; MCAM network; novel mask-guided class activation mapping network; CUHK03-NP; global depthwise convolution layer; person re-identification; loss function; feature information; mask-guided mapping loss; DukeMTMC-reID

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

References

    1. 1)
      • 5. Cai, H., Wang, Z., Cheng, J.: ‘Multi-scale body-part mask guided attention for person re-identification’, 2019, arXiv preprint arXiv:1904.11041.
    2. 2)
      • 7. Sun, Y., Zheng, L., Yang, Y., et al: ‘Beyond part models: person retrieval with refined part pooling (and a strong convolutional baseline)’. European Conf. on Computer Vision (ECCV), Munich, Germany, 2018, pp. 480–496.
    3. 3)
      • 9. Li, W., Zhu, X., Gong, S.: ‘Harmonious attention network for person re-identification’. Conf. on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, USA, 2018, pp. 2285–2294.
    4. 4)
      • 4. Song, C., Huang, Y., Ouyang, W., et al: ‘Mask-guided contrastive attention model for person re-identification’. Conf. on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, USA, 2018, pp. 1179–1188.
    5. 5)
      • 10. Wang, C., Zhang, Q., Huang, C., et al: ‘Mancs: a multi-task attentional network with curriculum sampling for person re-identification’. European Conf. on Computer Vision (ECCV), Munich, Germany, 2018, pp. 365–381.
    6. 6)
      • 6. Zhou, B., Khosla, A., Lapedriza, A., et al: ‘Learning deep features for discriminative localization’. Conf. on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016, pp. 2921–2929.
    7. 7)
      • 2. Zhao, H., Tian, M., Sun, S., et al: ‘Spindle net: person re-identification with human body region guided feature decomposition and fusion’. Conf. on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA, 2017, pp. 907–915.
    8. 8)
      • 1. Li, D., Chen, X., Zhang, Z., et al: ‘Learning deep context-aware features over body and latent parts for person re-identification’. Conf. on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA, 2017, pp. 7398–7407.
    9. 9)
      • 3. Kalayeh, M.M., Basaran, E., Gokmen, M., et al: ‘Human semantic parsing for person re-identification’. Conf. on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, USA, 2018, pp. 1062–1071.
    10. 10)
http://iet.metastore.ingenta.com/content/journals/10.1049/el.2020.1643
Loading

Related content

content/journals/10.1049/el.2020.1643
pub_keyword,iet_inspecKeyword,pub_concept
6
6
Loading