Re-ranking person re-identification using distance aggregation of k-nearest neighbors hierarchical tree
- PDF / 1,662,438 Bytes
- 24 Pages / 439.37 x 666.142 pts Page_size
- 53 Downloads / 205 Views
Re-ranking person re-identification using distance aggregation of k-nearest neighbors hierarchical tree Muhammad Hanif 1
1
1
1
& Hefei Ling & Weiyi Tian & Yuxuan Shi & Mudassar Rauf
2
Received: 26 December 2019 / Revised: 2 September 2020 / Accepted: 19 October 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Person re-identification is a challenging task due to the critical factors like illumination, occlusion, pose variation, view-points, low resolution, inter/intra-class variations, etc. Reidentification mainly treated as the target retrieval process, which can be improved by multi-query and re-ranking. Due to the high computational cost and consideration of fixed length gallery size, existing re-ranking approaches are not feasible for real-time reidentification applications, specifically with the variable-length gallery. We have proposed a fast yet effective re-ranking approach that utilize the pre-computed pair-wise distance used for initial ranking. We have integrated the advantages of nearest neighbors, hierarchical tree, and multi-query for re-ranking. We hypothesize that the hierarchy of knearest neighbors of an image leads to more positive matches in the child layer. Hence the aggregated distance decreases for true match and increases for false match images of the initial rank. We have structured k-nearest neighbor’s hierarchical tree and calculated the aggregated distance. Hierarchical nearest neighbors are treated as multi-query under distance aggregation. Final re-ranked distance is computed as the weighted sum of aggregated and actual distance. Our proposed re-ranking approach is computationally efficient, feasible for real-time applications, unsupervised, and completely automatic. The effectiveness of our proposed method (Source code isavailable upon request) has been verified by various experiments on MARS, Market-1501, DukeMTMC, and CUHK03 datasets. Keywords Re-ranking . Person re-identification . Distance aggregation . Hierarchical tree . knearest neighbors . Multi-query
* Muhammad Hanif [email protected]
1
School of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, China
2
State Key Laboratory of Digital Manufacturing Equipment and Technology, SANY Joint Lab of Advanced Manufacturing, Huazhong University of Science and Technology, Wuhan 430074, China
Multimedia Tools and Applications
1 Introduction Safety and security are always high priority concern for humans, which gradually improved from manual/human-based to semi-mechanized/sensors-based to fully-mechanized/vision-based [15, 25, 51]. As concerned with human surveillance, reidentification (ReID) of human is the most important and difficult task. Aim of the person ReID is to find a query/person of interest from the gallery/network of cameras and mainly treated as an image retrieval problem [8]. Person ReID has various applications, including security, surveillance, tracking, image retrieval, and forensic. It is one of the most challenging task
Data Loading...
