Research map / Continual learning / Domain and prompt-based
Lifelong Person Re-identification by Pseudo Task Knowledge Preservation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Apply PTKP instead of the strongest listed non-proposed baseline | Lifelong person Re-ID · Market-1501, DukeMTMC, CUHK-SYSU, and MSMT17 | mAP: 69.6 -> 77.0; 58.2 -> 63.8; 85.9 -> 88.4; 41.2 -> 41.9 | Helped |
| Apply PTKP instead of the strongest listed baseline | Lifelong person Re-ID generalization · CUHK01, CUHK03, GRID, SenseReID | Rank-1: 78.2 -> 79.1; 45.0 -> 54.1; 29.6 -> 31.5; 42.7 -> 44.8 | Helped |
| Add task-specific domain consistency loss (ablation) | PTKP · Market-1501, DukeMTMC, CUHK-SYSU, MSMT17 | mAP: Exp 1 -> Exp 2: 72.4 -> 74.2; 59.3 -> 62.5; 86.1 -> 86.2; 36.1 -> 37.1 | Helped |
| Add pseudo-task knowledge distillation (ablation) | PTKP with LDCL · Market-1501, DukeMTMC, CUHK-SYSU, MSMT17 | mAP: Exp 2 -> Exp 3: 74.2 -> 76.6; 62.5 -> 64.2; 86.2 -> 86.6; 37.1 -> 37.9 | Helped |
| Add pseudo-task identity discrimination (ablation) | PTKP with LDCL and LPT-KD · Market-1501, DukeMTMC, CUHK-SYSU, MSMT17 | mAP: Exp 3 -> Exp 4: 76.6 -> 76.5; 64.2 -> 63.5; 86.6 -> 88.2; 37.9 -> 42.1 | Mixed |
| Use feature-output distillation instead of pseudo-task knowledge distillation (ablation) | PTKP ablation · Market-1501, DukeMTMC, CUHK-SYSU, MSMT17 | mAP: Exp 4 vs Exp 5: 76.5 -> 73.7; 63.5 -> 62.1; 88.2 -> 86.4; 42.1 -> 41.2 | Mixed |
| Add ambiguity-aware instance weighting (ablation) | PTKP with Lbase, LDCL, LPT-KD, and LPT-ID · Market-1501, DukeMTMC, CUHK-SYSU, MSMT17 | mAP: Exp 4 -> Exp 7: 76.5 -> 77.0; 63.5 -> 63.8; 88.2 -> 88.4; 42.1 -> 41.9 | Mixed |
| Remove task-specific domain consistency loss (ablation) | PTKP · Market-1501, DukeMTMC, CUHK-SYSU, MSMT17 | mAP: Exp 7 vs Exp 6: 77.0 -> 74.6; 63.8 -> 60.8; 88.4 -> 88.3; 41.9 -> 40.1 | Mixed |
Closest papers by meaning
- Lifelong Person Re-Identification via Adaptive Knowledge Accumulation (CVPR 2021 · similarity 0.93)
- Lifelong Person Re-identification via Knowledge Refreshing and Consolidation (AAAI 2023 · similarity 0.92)
- LSTKC: Long Short-Term Knowledge Consolidation for Lifelong Person Re-identification (AAAI 2024 · similarity 0.92)
- Distribution-aware Knowledge Prototyping for Non-exemplar Lifelong Person Re-identification (CVPR 2024 · similarity 0.90)
- DKC: Differentiated Knowledge Consolidation for Cloth-Hybrid Lifelong Person Re-identification (CVPR 2025 · similarity 0.90)
- DASK: Distribution Rehearsing via Adaptive Style Kernel Learning for Exemplar-Free Lifelong Person Re-Identification (AAAI 2025 · similarity 0.89)
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.