Joint Pyramid Feature Representation Network for Vehicle Re-identification
- PDF / 2,029,040 Bytes
- 12 Pages / 595.224 x 790.955 pts Page_size
- 70 Downloads / 226 Views
Joint Pyramid Feature Representation Network for Vehicle Re-identification Xiangwei Lin1 · Huanqiang Zeng1
· Jinhui Hou1 · Jiuwen Cao2 · Jianqing Zhu3 · Jing Chen1
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Vehicle re-identification (Re-ID) technology plays an important role in the intelligent transportation system for smart city. Due to various uncertain factors in the real-world scenarios, (e.g., resolution variation, viewpoint variation, illumination changes, occlusion, etc., vehicle Re-ID is a very challenging task. To resist the adverse effect of resolution variation, a joint pyramid feature representation network (JPFRN) for vehicle Re-ID is proposed in this paper. Based on the consideration that various convolution blocks with different depths hold different resolutions and semantic information of the vehicle image, the proposed JPFRN method employs a base network to obtain multi-resolution vehicle features in the first stage. Then, a pyramid feature representation scheme is developed to reconstruct and integrate the obtained multi-resolution vehicle features together. Finally, these pyramid features are jointly represented for learning a more discriminative feature under the supervision of joint Triplet loss and softmax loss. Extensive experimental results on two commonly-used vehicle databases (i.e., VehicleID and VeRi) show that the proposed JPFRN is superior to multiple recently-developed vehicle Re-ID methods. Keywords Internet of Things · Intelligent transport system · Vehicle re-identification · Joint pyramid feature representation · Deep learning
1 Introduction In the era of smart city, the Internet of Things (IoT) technologies aim to provide the real-life physical objects with unique identifiers and ability to interact with each other without requiring human-to-human or human-to-computer interaction [1]. As an indispensable part of smart city, vehicle has been an important real-life physical object and received more and more attentions [2]. Due to the fact that the number of the vehicles is dramatically increased year by year, the intelligent management of the vehicles is highly required. For example, the number of vehicles on the road can be monitored in real time by a monitoring probe to control the passing time of the intersection; the images Huanqiang Zeng
[email protected] 1
School of Information Science and Engineering, Huaqiao University, Xiamen 361021, China
2
Artificial Intelligence Institute, Hangzhou Dianzi University, Hangzhou 310018, China
3
School of Engineering, Huaqiao University, Quanzhou 362021, China
of vehicles captured by various cameras can be used to automatically track the vehicle. The information interaction between these vehicles greatly reduces the waste of human and material resources, and thus significantly improves the efficiency of urban operation. Therefore, through the means of IoT, the Internet of Vehicles (IoV) has been emerging for autonomous industry and intelligent transportation system. For that, researchers from
