Disentangled feature learning network for vehicle re-identification

Computing and Communications

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Published

Standard

Disentangled feature learning network for vehicle re-identification. / Bai, Y.; Lou, Y.; Dai, Y. et al.
Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. ed. / Christian Bessiere. IJCAI, 2021. p. 474-480.

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Harvard

Bai, Y, Lou, Y, Dai, Y, Liu, J, Chen, Z & Duan, L-Y 2021, Disentangled feature learning network for vehicle re-identification. in C Bessiere (ed.), Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. IJCAI, pp. 474-480. <https://www.ijcai.org/proceedings/2020/66>

APA

Bai, Y., Lou, Y., Dai, Y., Liu, J., Chen, Z., & Duan, L.-Y. (2021). Disentangled feature learning network for vehicle re-identification. In C. Bessiere (Ed.), Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence (pp. 474-480). IJCAI. https://www.ijcai.org/proceedings/2020/66

Vancouver

Bai Y, Lou Y, Dai Y, Liu J, Chen Z, Duan LY. Disentangled feature learning network for vehicle re-identification. In Bessiere C, editor, Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. IJCAI. 2021. p. 474-480

Author

Bai, Y. ; Lou, Y. ; Dai, Y. et al. / Disentangled feature learning network for vehicle re-identification. Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. editor / Christian Bessiere. IJCAI, 2021. pp. 474-480

Bibtex

@inproceedings{91c42d37fc3f412db34dba741f98ced1,

title = "Disentangled feature learning network for vehicle re-identification",

abstract = "Vehicle Re-Identification (ReID) has attracted lots of research efforts due to its great significance to the public security. In vehicle ReID, we aim to learn features that are powerful in discriminating subtle differences between vehicles which are visually similar, and also robust against different orientations of the same vehicle. However, these two characteristics are hard to be encapsulated into a single feature representation simultaneously with unified supervision. Here we propose a Disentangled Feature Learning Network (DFLNet) to learn orientation specific and common features concurrently, which are discriminative at details and invariant to orientations, respectively. Moreover, to effectively use these two types of features for ReID, we further design a feature metric alignment scheme to ensure the consistency of the metric scales. The experiments show the effectiveness of our method that achieves state-of-the-art performance on three challenging datasets.",

author = "Y. Bai and Y. Lou and Y. Dai and Jun Liu and Z. Chen and L.-Y. Duan",

year = "2021",

month = jan,

day = "7",

language = "English",

pages = "474--480",

editor = "Christian Bessiere",

booktitle = "Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence",

publisher = "IJCAI",

}

RIS

TY - GEN

T1 - Disentangled feature learning network for vehicle re-identification

AU - Bai, Y.

AU - Lou, Y.

AU - Dai, Y.

AU - Liu, Jun

AU - Chen, Z.

AU - Duan, L.-Y.

PY - 2021/1/7

Y1 - 2021/1/7

N2 - Vehicle Re-Identification (ReID) has attracted lots of research efforts due to its great significance to the public security. In vehicle ReID, we aim to learn features that are powerful in discriminating subtle differences between vehicles which are visually similar, and also robust against different orientations of the same vehicle. However, these two characteristics are hard to be encapsulated into a single feature representation simultaneously with unified supervision. Here we propose a Disentangled Feature Learning Network (DFLNet) to learn orientation specific and common features concurrently, which are discriminative at details and invariant to orientations, respectively. Moreover, to effectively use these two types of features for ReID, we further design a feature metric alignment scheme to ensure the consistency of the metric scales. The experiments show the effectiveness of our method that achieves state-of-the-art performance on three challenging datasets.

AB - Vehicle Re-Identification (ReID) has attracted lots of research efforts due to its great significance to the public security. In vehicle ReID, we aim to learn features that are powerful in discriminating subtle differences between vehicles which are visually similar, and also robust against different orientations of the same vehicle. However, these two characteristics are hard to be encapsulated into a single feature representation simultaneously with unified supervision. Here we propose a Disentangled Feature Learning Network (DFLNet) to learn orientation specific and common features concurrently, which are discriminative at details and invariant to orientations, respectively. Moreover, to effectively use these two types of features for ReID, we further design a feature metric alignment scheme to ensure the consistency of the metric scales. The experiments show the effectiveness of our method that achieves state-of-the-art performance on three challenging datasets.

M3 - Conference contribution/Paper

SP - 474

EP - 480

BT - Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence

A2 - Bessiere, Christian

PB - IJCAI

ER -

Research

Links