Cross-Platform Video Person ReID: A New Benchmark Dataset and Adaptation Approach
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913490328879104 |
|---|---|
| author | Zhang, Shizhou Luo, Wenlong Cheng, De Yang, Qingchun Ran, Lingyan Xing, Yinghui Zhang, Yanning |
| author_facet | Zhang, Shizhou Luo, Wenlong Cheng, De Yang, Qingchun Ran, Lingyan Xing, Yinghui Zhang, Yanning |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_07500 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Cross-Platform Video Person ReID: A New Benchmark Dataset and Adaptation Approach Zhang, Shizhou Luo, Wenlong Cheng, De Yang, Qingchun Ran, Lingyan Xing, Yinghui Zhang, Yanning Computer Vision and Pattern Recognition 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. |
| title | Cross-Platform Video Person ReID: A New Benchmark Dataset and Adaptation Approach |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2408.07500 |