Re-ranking Person Re-identification with k-Reciprocal Encoding

TLDR

Re-ranking is essential for improving person re-identification accuracy, yet the community has largely neglected fully automatic, unsupervised re-ranking methods. The study proposes a k‑reciprocal encoding approach that re‑ranks re‑ID results by assuming that gallery images sharing k‑reciprocal nearest neighbors with a probe are more likely true matches. The method encodes each image’s k‑reciprocal nearest neighbors into a vector, uses the Jaccard distance for re‑ranking, and combines it with the original distance, all without requiring labeled data or human input. Experiments on Market‑1501, CUHK03, MARS, and PRW datasets demonstrate the method’s effectiveness.

Abstract

When considering person re-identification (re-ID) as a retrieval process, re-ranking is a critical step to improve its accuracy. Yet in the re-ID community, limited effort has been devoted to re-ranking, especially those fully automatic, unsupervised solutions. In this paper, we propose a k-reciprocal encoding method to re-rank the re-ID results. Our hypothesis is that if a gallery image is similar to the probe in the k-reciprocal nearest neighbors, it is more likely to be a true match. Specifically, given an image, a k-reciprocal feature is calculated by encoding its k-reciprocal nearest neighbors into a single vector, which is used for re-ranking under the Jaccard distance. The final distance is computed as the combination of the original distance and the Jaccard distance. Our re-ranking method does not require any human interaction or any labeled data, so it is applicable to large-scale datasets. Experiments on the large-scale Market-1501, CUHK03, MARS, and PRW datasets confirm the effectiveness of our method.