AS-RIG: Adaptive Selection of Reconstructed Input by Generator or Interpolation for Person Re-Identification in Cross-Modality Visible and Thermal Images

Jin Kyu Kang,Kang Ryoung Park,Min Beom Lee,Hyo Sik Yoon

doi:10.1109/access.2021.3051637

Jin Kyu Kang, Kang Ryoung Park + Show 2 more

Open Access

https://doi.org/10.1109/access.2021.3051637

Copy DOI

Abstract

Multimodal camera-based person re-identification (ReID) is important in the field of intelligent surveillance. Thermal cameras can solve the problem in that visible-light cameras cannot acquire the valid feature information of a person under poor illumination conditions. However, thermal cameras usually have lower frame resolution than visible-light cameras. To overcome this problem, we propose an adaptive selection of reconstructed input by generator or interpolation (AS-RIG) method, which can adaptively select the generative adversarial network (GAN), or an interpolation method (bi-linear or bi-cubic). AS-RIG automatically selects a resolution-model using the mean-squared error (MSE), feature distance (FD), and structural similarity (SSIM). To verify the performance of our proposed method, two open databases are used: the DBPerson-Recog-DB1 and Sun Yat-set University multiple modality Re-ID (SYSU-MM01). Infrared frames from both databases are resized to be smaller than the original ones for experimentation. Experimental results show that our generator outperforms traditional interpolation methods. In addition, the person ReID experimental results demonstrate that AS-RIG outperforms non-adaptive selection methods and state-of-the-art methods.

Highlights

Person re-identification (ReID) aims to match a specific person having varying viewpoints and poses from two or more frames that are captured from more than one camera
In this work, we proposed a new perspective for improving the performance of cross-modality person ReID using AS-RIG
In order to improve the cross-modality person ReID using the traditional interpolation method, reconstruction by generator was applied as input data reconstruction

Summary

Introduction

Person re-identification (ReID) aims to match a specific person having varying viewpoints and poses from two or more frames that are captured from more than one camera. Compared to the tracking algorithm, this is difficult because it is an environment that has non-continuity with respect to the axis in time. Research on person ReID has been performed owing to the need for intelligent surveillance systems [1]. Research to re-identify a specific person using different visible-light cameras (CCTV) in the daytime environment is the main focus [2]–[4]. Visiblelight cameras are vulnerable to illumination conditions.

Objectives

Methods

Results

Conclusion