Person Re-identification on MSMT17
0.882mAPCLIP-SCGI
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CLIP-SCGIBackbone=ViT, Resolution=256x1282024.10 | 0.882 | 0.929 | — | — | — | — | |
| CLIP-SCGIBackbone=ViT-B/16, Modality=Image-based, Supervision=External Supervision2026.04 | 0.882 | 0.929 | — | — | — | — | |
| PromptSGBackbone=ViT, Resolution=256x1282024.10 | 0.872 | 0.926 | — | — | — | — | |
| PromptSG2026.03 | 0.872 | 0.926 | — | — | — | — | |
| PromptSGBackbone=ViT-B/16, Modality=Image-based, Supervision=No External Supervision2026.04 | 0.872 | 0.926 | — | — | — | — | |
| GCPBase Method=PHA2025.10 | 0.834 | 0.897 | — | — | — | — | |
| ABDNet2025.10 | 0.823 | 0.608 | — | — | — | — | |
| GCPBase Method=TransReID2025.10 | 0.821 | 0.854 | — | — | — | — | |
| α-FPSBase Method=PHA2025.10 | 0.815 | 0.854 | — | — | — | — | |
| FlipReIDBackbone=ResNeSt50, Inference mode=double images, Flipping loss=with, Re-ranking=true2021.05 | 0.813 | 0.875 | — | — | — | — | |
| k-centroidBase Method=PHA2025.10 | 0.811 | 0.851 | — | — | — | — | |
| CLIP-SCGIBackbone=CNN, Resolution=256x1282024.10 | 0.809 | 0.879 | — | — | — | — | |
| PersonViTBackbone=ViT-B/162024.08 | 0.808 | 0.92 | — | — | — | — | |
| FlipReIDBackbone=IBN-ResNet50, Inference mode=double images, Flipping loss=with, Re-ranking=true2021.05 | 0.802 | 0.873 | — | — | — | — | |
| PersonMAEBackbone=ViT-B/162024.08 | 0.798 | 0.914 | — | — | — | — | |
| PersonMAEBackbone=ViT-B2026.01 | 0.798 | 0.914 | — | — | — | — | |
| FlipReIDBackbone=ResNet50, Inference mode=double images, Flipping loss=with, Re-ranking=true2021.05 | 0.795 | 0.863 | — | — | — | — | |
| HPLVenue=AAAI’262026.02 | 0.79 | 0.91 | — | — | — | — | |
| HPLCategory=Unified2026.06 | 0.79 | 0.91 | — | — | — | — | |
| CLASP (TransReID)Backbone=Swin-B2026.01 | 0.788 | 0.918 | — | — | — | — | |
| DRFormer2026.02 | 0.787 | 0.913 | — | — | — | — | |
| ReTextCategory=Unified2026.06 | 0.787 | 0.914 | — | — | — | — | |
| SAGA+CLIMB (Ours)Backbone=ViT-B/16, Modality=Image-based, Supervision=No External Supervision2026.04 | 0.784 | 0.9 | — | — | — | — | |
| HAPBackbone=ViT-B2026.01 | 0.781 | — | — | — | — | — | |
| SOLDIERArch=Swin-Base, Train Data=LU4M+MSMT172025.04 | 0.7801 | 0.9112 | — | — | — | — | |
| SAGA-ReID (Ours)Backbone=ViT-B/16, Modality=Image-based, Supervision=No External Supervision2026.04 | 0.774 | 0.905 | — | — | — | — | |
| CLASP (TransReID)Backbone=Swin-S2026.01 | 0.772 | 0.912 | — | — | — | — | |
| SOLIDERBackbone=Swin-B2023.03 | 0.771 | 0.907 | — | — | — | — | |
| SOLIDERBackbone=Swin-B2024.08 | 0.771 | 0.907 | — | — | — | — | |
| SOLIDERBackbone=Swin-B2026.01 | 0.771 | 0.907 | — | — | — | — | |
| SOLIDER2025.10 | 0.771 | 0.907 | — | — | — | — | |
| FastReIDBackbone=ResNet101-ibn, Query Expansion=true2020.06 | 0.769 | 0.879 | — | — | — | — | |
| SOLIDERBackbone=Swin-S2023.03 | 0.769 | 0.908 | — | — | — | — | |
| CLIMB-ReIDBackbone=ViT-B/16, Modality=Video-based, Supervision=No External Supervision2026.04 | 0.769 | 0.896 | — | — | — | — | |
| OursCategory=Unified2026.06 | 0.769 | 0.901 | — | — | — | — | |
| Adaptive L2 RegularizationBackbone=ResNet152, Re-ranking=applied2020.07 | 0.767 | 0.849 | — | — | — | — | |
| DynaMixCategory=I2I-only2026.06 | 0.763 | 0.904 | — | — | — | — | |
| HAPArch=ViT-Base, Train Data=LU4M+MSMT172025.04 | 0.7585 | 0.8913 | — | — | — | — | |
| CLIP-ReIDBackbone=ViT, Side Information Embedding (SIE)=true, Overlapping Patch (OLP)=true2022.11 | 0.758 | 0.897 | — | — | — | — | |
| CLIP-ReIDBackbone=ViT-B/16, Modality=Image-based, Supervision=No External Supervision2026.04 | 0.758 | 0.897 | — | — | — | — | |
| PersonMAEBackbone=ViT-S/162024.08 | 0.752 | 0.891 | — | — | — | — | |
| PASS2023.03 | 0.743 | 0.897 | — | — | — | — | |
| PersonViTBackbone=ViT-S/162024.08 | 0.743 | 0.892 | — | — | — | — | |
| PASSBackbone=ViT-B2026.01 | 0.743 | 0.897 | — | — | — | — | |
| TransReID + Pose2IDweights=official pre-trained2025.03 | 0.7406 | 0.8655 | — | — | — | — | |
| CLIP-ReIDBackbone=ViT, Resolution=256x1282024.10 | 0.734 | 0.887 | — | — | — | — | |
| CLIP-ReIDVenue=AAAI’232026.02 | 0.734 | 0.887 | — | — | — | — | |
| CLIP-ReIDBackbone=ViT2022.11 | 0.734 | 0.887 | — | — | — | — | |
| CLIP-ReID2026.03 | 0.734 | 0.887 | — | — | — | — | |
| CLIP-ReIDBackbone=ViT-B/16, Method Category=CLIP-based method, Ref.=AAAI’232026.03 | 0.734 | 0.887 | — | — | — | — | |
| CLIP-ReIDCategory=I2I-only2026.06 | 0.734 | 0.887 | — | — | — | — | |
| IRM2025.10 | 0.724 | 0.869 | — | — | — | — | |
| IRMCategory=Unified2026.06 | 0.724 | 0.869 | — | — | — | — | |
| MUVA2026.03 | 0.722 | 0.89 | — | — | — | — | |
| DiPSize=384, Stride=122022.12 | 0.718 | 0.873 | — | — | — | — | |
| PASSBackbone=ViT-B/162024.08 | 0.718 | 0.882 | — | — | — | — | |
| PASSCategory=I2I-only2026.06 | 0.718 | 0.882 | — | — | — | — | |
| PartFormer*Backbone=ViT-Base, Sliding window setting=true2024.08 | 0.713 | 0.871 | — | — | — | — | |
| Pose-dIVEBase Model=CLIP-reID2024.06 | 0.71 | 0.875 | — | — | — | — | |
| PersonViT*Backbone=ViT-S/162024.08 | 0.709 | 0.873 | — | — | — | — | |
| DC-FormerVenue=AAAI’232026.02 | 0.707 | 0.869 | — | — | — | — | |
| DiPSize=256, Stride=122022.12 | 0.706 | 0.863 | — | — | — | — | |
| DiPSize=384, Stride=162022.12 | 0.706 | 0.866 | — | — | — | — | |
| TMGF (w/ GT)Supervision Type=Fully supervised, Backbone=Transformer, Reference=This work2022.11 | 0.703 | 0.882 | 0.941 | 0.954 | — | — | |
| TransReIDBackbone=ViT-B/16, Size=384x128, Sliding-window setting=true2021.02 | 0.694 | 0.862 | — | — | — | — | |
| TransReIDSupervision Type=Fully supervised, Backbone=Transformer, Reference=ICCV212022.11 | 0.694 | 0.862 | — | — | — | — | |
| TransReIDSize=384, Stride=122022.12 | 0.694 | 0.862 | — | — | — | — | |
| TransReID2023.03 | 0.694 | 0.862 | — | — | — | — | |
| TransReIDBackbone=ViT-B2026.01 | 0.694 | 0.862 | — | — | — | — | |
| TransReIDBackbone=ViT, Side Information Embedding (SIE)=true, Overlapping Patch (OLP)=true, Input Resolution=> 256x1282022.11 | 0.694 | 0.862 | — | — | — | — | |
| TransReID2024.06 | 0.694 | 0.862 | — | — | — | — | |
| PASSBackbone=ViT-S/162024.08 | 0.691 | 0.865 | — | — | — | — | |
| PATHBackbone=ViT-B2026.01 | 0.691 | — | — | — | — | — | |
| PHA*Backbone=ViT-Base, Sliding window setting=true2024.08 | 0.689 | 0.861 | — | — | — | — | |
| PHABackbone=ViT, Resolution=256x1282024.10 | 0.689 | 0.861 | — | — | — | — | |
| PHAVenue=CVPR’232026.02 | 0.689 | 0.861 | — | — | — | — | |
| PHABackbone=ViT-B/16, Method Category=ViT-based method, Ref.=CVPR’232026.03 | 0.689 | 0.861 | — | — | — | — | |
| PHA2025.10 | 0.689 | 0.861 | — | — | — | — | |
| ALADIN (Ours)Backbone=ViT-B/16, Method Category=CLIP-based method2026.03 | 0.688 | 0.865 | — | — | — | — | |
| PromptSGBackbone=CNN, Resolution=256x1282024.10 | 0.685 | 0.86 | — | — | — | — | |
| PartFormerBackbone=ViT-Base2024.08 | 0.683 | 0.852 | — | — | — | — | |
| Pose-dIVEBase Model=SOLIDER2024.06 | 0.683 | 0.859 | — | — | — | — | |
| TransReID-SSL (w/ GT)Supervision Type=Fully supervised, Backbone=Transformer, Reference=arxiv212022.11 | 0.681 | 0.861 | — | — | — | — | |
| CLASP (TransReID)Backbone=Swin-T2026.01 | 0.681 | 0.868 | — | — | — | — | |
| FlipReIDBackbone=ResNeSt50, Inference mode=double images, Flipping loss=with, Re-ranking=false2021.05 | 0.68 | 0.856 | — | — | — | — | |
| PNLPre-training method=PNL2022.03 | 0.68 | 0.86 | — | — | — | — | |
| FlipReID2024.06 | 0.68 | 0.856 | — | — | — | — | |
| CLIP-reID2024.06 | 0.68 | 0.858 | — | — | — | — | |
| TransReIDweights=official pre-trained2025.03 | 0.678 | 0.8533 | — | — | — | — | |
| FlipReIDBackbone=ResNeSt50, Inference mode=double images, Flipping loss=without, Re-ranking=false2021.05 | 0.676 | 0.853 | — | — | — | — | |
| DiPSize=256, Stride=162022.12 | 0.675 | 0.846 | — | — | — | — | |
| FlexiReIDVenue=ICML’252026.02 | 0.675 | 0.837 | — | — | — | — | |
| TransReIDBackbone=ViT-B/16, Size=256x128, Sliding-window setting=true2021.02 | 0.674 | 0.853 | — | — | — | — | |
| TransReIDSize=256, Stride=122022.12 | 0.674 | 0.853 | — | — | — | — | |
| SOLIDERBackbone=Swin-T2023.03 | 0.674 | 0.859 | — | — | — | — | |
| TransReID*Backbone=ViT-Base, Sliding window setting=true2024.08 | 0.674 | 0.853 | — | — | — | — | |
| TransReIDBackbone=ViT, Resolution=256x1282024.10 | 0.674 | 0.853 | — | — | — | — | |
| TransReIDVenue=ICCV'212026.02 | 0.674 | 0.853 | — | — | — | — | |
| TransReIDBackbone=ViT, Side Information Embedding (SIE)=true, Overlapping Patch (OLP)=true2022.11 | 0.674 | 0.853 | — | — | — | — | |
| TransReIDInput image size=256x1282026.03 | 0.674 | 0.853 | — | — | — | — |