Soft Person Reidentification Network Pruning via Blockwise Adjacent Filter Decaying

Page view(s)
65
Checked on Jul 11, 2024
Soft Person Reidentification Network Pruning via Blockwise Adjacent Filter Decaying
Title:
Soft Person Reidentification Network Pruning via Blockwise Adjacent Filter Decaying
Journal Title:
IEEE Transactions on Cybernetics
Publication Date:
15 December 2021
Citation:
Wang, X., Zheng, Z., He, Y., Yan, F., Zeng, Z., & Yang, Y. (2021). Soft Person Reidentification Network Pruning via Blockwise Adjacent Filter Decaying. IEEE Transactions on Cybernetics, 1–15. https://doi.org/10.1109/tcyb.2021.3130047
Abstract:
Deep learning has shown significant successes in person reidentification (re-id) tasks. However, most existing works focus on discriminative feature learning and impose complex neural networks, suffering from low inference efficiency. In fact, feature extraction time is also crucial for real-world applications and lightweight models are needed. Prevailing pruning methods usually pay attention to compact classification models. However, these methods are suboptimal for compacting re-id models, which usually produce continuous features and are sensitive to network pruning. The key point of pruning re-id models is how to retain the original filter distribution in continuous features as much as possible. In this work, we propose a blockwise adjacent filter decaying method to fill this gap. Specifically, given a trained model, we first evaluate the redundancy of filters based on the adjacency relationships to preserve the original filter distribution. Second, previous layerwise pruning methods ignore that discriminative information is enhanced block-by-block. Therefore, we propose a blockwise filter pruning strategy to better utilize the block relations in the pretrained model. Third, we propose a novel filter decaying policy to progressively reduce the scale of redundant filters. Different from conventional soft filter pruning that directly sets the filter values as zeros, the proposed filter decaying can keep the pretrained knowledge as much as possible. We evaluate our method on three popular person reidentification datasets, that is: 1) Market-1501; 2) DukeMTMC-reID; and 3) MSMT17_V1. The proposed method shows superior performance to the existing state-of-the-art pruning methods. After pruning over 91.9% parameters on DukeMTMC-reID, the Rank-1 accuracy only drops 3.7%, demonstrating its effectiveness for compacting person reidentification.
License type:
Publisher Copyright
Funding Info:
This research is supported by core funding from: A*STAR SERC Central Research Fund
Grant Reference no. : NA

This work was supported in part by China Scholarship Council under Grant 201908350025; in part by the National Natural Science Foundation of China under Grant 61871464; in part by the National Natural Science Foundation of Fujian Province under Grant 2020J01266 and Grant 2021J011186; in part by the “Climbing” Program of XMUT under Grant XPDKT20031; and in part by the Program of XMUT for High-Level Talents Introduction Plan under Grant YKJ19003R.
Description:
© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works
ISSN:
2168-2267
2168-2275
Files uploaded:

File Size Format Action
soft-reid.pdf 2.82 MB PDF Open