Research map / Label-efficient and robust learning / Recognition and long-tail
VIPerson: Flexibly Generating Virtual Identity for Person Re-Identification
Xiao-Wen Zhang, Delong Zhang, Yi-Xing Peng, Zhi Ouyang, Jingke Meng, Wei-Shi Zheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add camera-realistic style realization (ablation) | VIPerson · Market1501 direct transfer | rank-1: 33.6 -> 66.5 | Improved |
| add camera-realistic style realization (ablation) | VIPerson · MSMT17 direct transfer | rank-1: 7.5 -> 32.5 | Improved |
| add scene variations (segment-and-paste) (ablation) | VIPerson · MSMT17 direct transfer | rank-1: 0.9 -> 7.5 | Improved |
| add scene variations (segment-and-paste) (ablation) | VIPerson · Market1501 direct transfer | rank-1: 11.7 -> 33.6 | Improved |
| add pose variations (ablation) | VIPerson · MSMT17 direct transfer | rank-1: 20.5 -> 32.5 | Improved |
| add pose variations (ablation) | VIPerson · Market1501 direct transfer | rank-1: 39.8 -> 66.5 | Improved |
| add hard identities via text embedding dropout (ablation) | VIPerson · MSMT17 and Market1501 (5000 identities) | rank-1 | Improved |
| increase number of identities from 1000 to 3000, then further (ablation) | VIPerson · MSMT17 and Market1501 | rank-1 | Mixed |
| increase intra-ID group scale (ablation) | VIPerson · MSMT17 and Market1501 | rank-1 | Improved |
| use VIPerson* with target-scenario pose, scene and style information | VIPerson · MSMT17 direct transfer | rank-1: 32.5 -> 34.3 | Improved |
| use VIPerson instead of UnrealPerson for direct transfer | UnrealPerson · MSMT17 direct transfer | rank-1: 26.8 -> 32.5 | Improved |
| pre-train on VIPerson instead of UnrealPerson before fine-tuning | UnrealPerson · MSMT17 supervised fine-tuning | rank-1: 74.5 -> 76.5 | Improved |
Closest papers by meaning
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.90)
- Diffusion-based Synthetic Data Generation for Visible-Infrared Person Re-Identification (AAAI 2025 · similarity 0.88)
- When Person Re-Identification Meets Event Camera: A Benchmark Dataset and an Attribute-Guided Re-Identification Framework (AAAI 2026 · similarity 0.88)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · similarity 0.87)
- Alice Benchmarks: Connecting Real World Re-Identification with the Synthetic (ICLR 2024 · similarity 0.86)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
