An Improved AlignedReID Method for Vehicle Re-identification

EasyChair Preprint 1422

Abstract

In this paper, we propose an improved vehicle re-identification method based on the combination between the AlignedReID and the Stochastic Weight Averaging (SWA). AlignedReID extracts a global feature and local features of a vehicle's image and performs joint learning. Local automatic alignment is achieved by computing the shortest path between the two sets of local features, so that global feature learning can benefit from local feature learning. By running an optimizer with a high constant learning rate, the SWA averages the weight of the model to ensure that a better weight combination can be found. Our improved method surpasses the most advanced methods on the VehicleID dataset and VeRi-776 dataset. In order to better solve the task of vehicle re-identification in residential area, we have made the Oeasy-Parking dataset and experimented with our methods, and achieved good results.

Keyphrases: AlignedReID, SWA, Vehicle re-identification

@booklet{EasyChair:1422,
  author    = {Yu Deng and Jinhui Xu and Shuo Chen and Yong Song},
  title     = {An Improved AlignedReID Method for Vehicle Re-identification},
  howpublished = {EasyChair Preprint 1422},
  year      = {EasyChair, 2019}}