Deep attention aware feature learning for person re-Identification
Deep attention aware feature learning for person re-Identification
复制标题
用于人员重新识别的深度注意力感知特征学习
DOI：
10.1016/j.patcog.2022.108567
复制
发表时间：
2022-02-06
影响因子：
8
通讯作者：
Zeng, Hui
中科院分区：
文献类型：
--
作者：
Chen, Yifan;Wang, Han;Zeng, Hui
A B S T R A C T视觉注意力已被证明是有效的，在提高性能的人重新识别。大多数现有的方法通过学习额外的注意力图来重新加权特征图以重新识别人，从而直观地应用视觉注意力，然而，这种方法不可避免地增加了模型复杂度和推理时间。在本文中，我们建议在不改变原始结构的情况下，将预测注意力地图的能力作为个人ReID网络的额外目标，从而保持相同的推理时间和模型大小。两种类型的注意力地图已被认为是使学习的特征地图分别意识到的人和相关的身体部位。在全局上，提出了一个整体注意力分支（HAB），使骨干提取的特征图能够集中在人身上，从而减轻背景的影响。在局部，提出了部分注意力分支（PAB），将提取的特征解耦为若干组，分别负责不同的身体部位，从而提高了对姿态变化和部分遮挡的鲁棒性。这两种注意力是通用的，可以并入现有的ReID网络。我们已经在两个典型的网络（TriNet [1]和Bag of Tricks [2]）上测试了它的性能，并在五个广泛使用的数据集上观察到显着的性能改进。(c)2022爱思唯尔有限公司版权所有。
A B S T R A C T Visual attention has proven to be effective in improving the performance of person re-identification. Most existing methods apply visual attention heuristically by learning an additional attention map to re-weight the feature maps for person re-identification, however, this kind of methods inevitably increase the model complexity and inference time. In this paper, we propose to incorporate the ability of predicting attention maps as additional objectives in a person ReID network without changing the original structure, thus maintain the same inference time and model size. Two kinds of attention maps have been considered to make the learned feature maps being aware of the person and related body parts respectively. Globally, a holistic attention branch (HAB) is proposed to make the feature maps obtained by backbone could focus on persons so as to alleviate the influence of background. Locally, a partial attention branch (PAB) is proposed to make the extracted features can be decoupled into several groups that are separately responsible for different body parts, thus increasing the robustness to pose variation and partial occlusion. These two kinds of attentions are universal and can be incorporated into existing ReID networks. We have tested its performance on two typical networks (TriNet [1] and Bag of Tricks [2]) and observed significant performance improvement on five widely used datasets. (c) 2022 Elsevier Ltd. All rights reserved.