Pedestrian Retrieval Using Valuable
Absence Augmentation
Xiaolong Hao
1,2 , Shuang Liu
1,2(B) , Zhong Zhang
1,2 , and Tariq S. Durrani
3
1 Tianjin Key Laboratory of Wireless Mobile Communications
and Power Transmission, Tianjin Normal University, Tianjin, China
shuangliu.tjnu@gmail.com
2 College of Electronic and Communication Engineering,
Tianjin Normal University, Tianjin, China
haoxiaolong17@gmail.com, zhang8848@gmail.com
3 Department of Electronic and Electrical Engineering,
University of Strathclyde, Glasgow, Scotland, UK
t.durrani@strath.ac.uk
Abstract. In this paper, we propose a novel data augmentation method
named valuable absence augmentation (VAA) in order to alleviate the
overfitting and evaluate the influence of the pedestrian valuable parts for
the network performance. Specifically, we first train a base convolutional
neural network model and obtain the attention map of the pedestrian.
Then, we use the attention map to generate new samples. Finally, original samples and new samples are combined to fine-tune the base network model. We conduct experiments on a large-scale pedestrian retrieval
database, i.e., Market-1501. Experimental results show that the pedestrian valuable part has a crucial influence for the network performance
and that the proposed method achieves better performance than other
state-of-the-art methods.
Keywords: Convolutional neural networks · Pedestrian retrieval ·
Data augmentation
1 Introduction
Pedestrian retrieval aims at retrieving a specified pedestrian from a large-scale
image pool (Gallery) that consists of images taken by different cameras at different times [1,2]. It is widely used in many applications including crowd counting,
multi-camera tracking, video surveillance, etc. [3–6], and gradually becomes a
hot research focus. Although pedestrian retrieval has been studied for many
years, it still confronts many challenges such as background clutter, different
illumination, various body poses, and so on.
With the help of the convolutional neural networks (CNNs), pedestrian
retrieval has achieved remarkable performance [7–10]. However, when an excesc
Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 254–261, 2020.
https://doi.org/10.1007/978-981-15-0187-6_29
Absence Augmentation
Xiaolong Hao
1,2 , Shuang Liu
1,2(B) , Zhong Zhang
1,2 , and Tariq S. Durrani
3
1 Tianjin Key Laboratory of Wireless Mobile Communications
and Power Transmission, Tianjin Normal University, Tianjin, China
shuangliu.tjnu@gmail.com
2 College of Electronic and Communication Engineering,
Tianjin Normal University, Tianjin, China
haoxiaolong17@gmail.com, zhang8848@gmail.com
3 Department of Electronic and Electrical Engineering,
University of Strathclyde, Glasgow, Scotland, UK
t.durrani@strath.ac.uk
Abstract. In this paper, we propose a novel data augmentation method
named valuable absence augmentation (VAA) in order to alleviate the
overfitting and evaluate the influence of the pedestrian valuable parts for
the network performance. Specifically, we first train a base convolutional
neural network model and obtain the attention map of the pedestrian.
Then, we use the attention map to generate new samples. Finally, original samples and new samples are combined to fine-tune the base network model. We conduct experiments on a large-scale pedestrian retrieval
database, i.e., Market-1501. Experimental results show that the pedestrian valuable part has a crucial influence for the network performance
and that the proposed method achieves better performance than other
state-of-the-art methods.
Keywords: Convolutional neural networks · Pedestrian retrieval ·
Data augmentation
1 Introduction
Pedestrian retrieval aims at retrieving a specified pedestrian from a large-scale
image pool (Gallery) that consists of images taken by different cameras at different times [1,2]. It is widely used in many applications including crowd counting,
multi-camera tracking, video surveillance, etc. [3–6], and gradually becomes a
hot research focus. Although pedestrian retrieval has been studied for many
years, it still confronts many challenges such as background clutter, different
illumination, various body poses, and so on.
With the help of the convolutional neural networks (CNNs), pedestrian
retrieval has achieved remarkable performance [7–10]. However, when an excesc
Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 254–261, 2020.
https://doi.org/10.1007/978-981-15-0187-6_29