Cross-Platform Video Person ReID: A New Benchmark Dataset and Adaptation Approach

TL;DR

This paper introduces a large-scale Ground-to-Aerial Video person ReID dataset and a novel cross-platform ReID method using vision-language models and adaptive prompts, improving performance across diverse scenarios.

Contribution

It presents the first Ground-to-Aerial video ReID dataset and a new adaptation approach leveraging CLIP and platform-bridge prompts for cross-platform alignment.

Findings

Proposed G2A-VReID dataset with 185,907 images and 2,788 identities.

The VSLA-CLIP method outperforms existing video ReID techniques.

Effective visual feature alignment across different platforms.

Abstract

In this paper, we construct a large-scale benchmark dataset for Ground-to-Aerial Video-based person Re-Identification, named G2A-VReID, which comprises 185,907 images and 5,576 tracklets, featuring 2,788 distinct identities. To our knowledge, this is the first dataset for video ReID under Ground-to-Aerial scenarios. G2A-VReID dataset has the following characteristics: 1) Drastic view changes; 2) Large number of annotated identities; 3) Rich outdoor scenarios; 4) Huge difference in resolution. Additionally, we propose a new benchmark approach for cross-platform ReID by transforming the cross-platform visual alignment problem into visual-semantic alignment through vision-language model (i.e., CLIP) and applying a parameter-efficient Video Set-Level-Adapter module to adapt image-based foundation model to video ReID tasks, termed VSLA-CLIP. Besides, to further reduce the great discrepancy across the platforms, we also devise the platform-bridge prompts for efficient visual feature alignment. Extensive experiments demonstrate the superiority of the proposed method on all existing video ReID datasets and our proposed G2A-VReID dataset.