# Learning Resolution-Invariant Deep Representations for Person Re-Identification

## TL;DR

This paper introduces RAIN, a novel end-to-end network that learns resolution-invariant features for person re-identification, effectively handling cross-resolution mismatches in real-world scenarios.

## Contribution

The paper proposes a new adversarial learning-based architecture, RAIN, that extracts resolution-invariant features for person re-ID without relying on separate super-resolution models.

## Key findings

- The model effectively recognizes low-resolution images unseen during training.
- RAIN outperforms existing methods in cross-resolution person re-ID tasks.
- Extension to semi-supervised learning demonstrates scalability for real-world applications.

## Abstract

Person re-identification (re-ID) solves the task of matching images across cameras and is among the research topics in vision community. Since query images in real-world scenarios might suffer from resolution loss, how to solve the resolution mismatch problem during person re-ID becomes a practical problem. Instead of applying separate image super-resolution models, we propose a novel network architecture of Resolution Adaptation and re-Identification Network (RAIN) to solve cross-resolution person re-ID. Advancing the strategy of adversarial learning, we aim at extracting resolution-invariant representations for re-ID, while the proposed model is learned in an end-to-end training fashion. Our experiments confirm that the use of our model can recognize low-resolution query images, even if the resolution is not seen during training. Moreover, the extension of our model for semi-supervised re-ID further confirms the scalability of our proposed method for real-world scenarios and applications.