Improving Person Re-identification with Iterative Impression Aggregation

TL;DR

This paper introduces an iterative impression aggregation method for person re-identification that updates probe representations with new gallery information, improving accuracy and achieving state-of-the-art results.

Contribution

It proposes a simple attentional aggregation approach that iteratively refines person representations, enhancing baseline models and enabling integration of various features and metrics.

Findings

Achieves competitive results on CUHK03, Market-1501, and DukeMTMC datasets.

Improves baseline re-ID performance and matches advanced re-ranking methods.

Attains state-of-the-art results with stronger representations and metrics.

Abstract

Our impression about one person often updates after we see more aspects of him/her and this process keeps iterating given more meetings. We formulate such an intuition into the problem of person re-identification (re-ID), where the representation of a query (probe) image is iteratively updated with new information from the candidates in the gallery. Specifically, we propose a simple attentional aggregation formulation to instantiate this idea and showcase that such a pipeline achieves competitive performance on standard benchmarks including CUHK03, Market-1501 and DukeMTMC. Not only does such a simple method improve the performance of the baseline models, it also achieves comparable performance with latest advanced re-ranking methods. Another advantage of this proposal is its flexibility to incorporate different representations and similarity metrics. By utilizing stronger representations and metrics, we further demonstrate state-of-the-art person re-ID performance, which also validates the general applicability of the proposed method.