Image Re-Identification: Where Self-supervision Meets Vision-Language Learning

TL;DR

This paper introduces SVLL-ReID, a novel two-stage training method that combines self-supervision with pre-trained CLIP to improve image re-identification performance without relying on explicit text labels.

Contribution

It is the first to integrate self-supervision with CLIP for image ReID through a two-stage training process, enhancing discriminability of features and prompts.

Findings

SVLL-ReID outperforms state-of-the-art methods on six benchmarks.

Language self-supervision improves text prompt distinguishability.

Vision self-supervision enhances image feature discriminability.

Abstract

Recently, large-scale vision-language pre-trained models like CLIP have shown impressive performance in image re-identification (ReID). In this work, we explore whether self-supervision can aid in the use of CLIP for image ReID tasks. Specifically, we propose SVLL-ReID, the first attempt to integrate self-supervision and pre-trained CLIP via two training stages to facilitate the image ReID. We observe that: 1) incorporating language self-supervision in the first training stage can make the learnable text prompts more distinguishable, and 2) incorporating vision self-supervision in the second training stage can make the image features learned by the image encoder more discriminative. These observations imply that: 1) the text prompt learning in the first stage can benefit from the language self-supervision, and 2) the image feature learning in the second stage can benefit from the vision self-supervision. These benefits jointly facilitate the performance gain of the proposed SVLL-ReID. By conducting experiments on six image ReID benchmark datasets without any concrete text labels, we find that the proposed SVLL-ReID achieves the overall best performances compared with state-of-the-arts. Codes will be publicly available at https://github.com/BinWangGzhu/SVLL-ReID.