Text-based Person Search on CUHK-PEDES
77.43Recall@1DAPL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DAPL2024.05 | 77.43 | 90.73 | 94.2 | 68.35 | |
| CONQUERImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 77.13 | 90.06 | — | 68.75 | |
| APTM2023.06 | 76.53 | 90.04 | 94.15 | 66.91 | |
| APTMVenue=MM’232024.05 | 76.53 | 90.04 | 94.15 | 66.91 | |
| RaSaVenue=IJCAI’232024.05 | 76.51 | 90.29 | 94.25 | 69.38 | |
| RDEImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 75.94 | 90.14 | — | 67.56 | |
| RDEVenue=CVPR’242024.05 | 75.94 | 90.14 | 94.12 | 67.56 | |
| DPImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 75.66 | 90.59 | — | 66.58 | |
| CFAMImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 75.6 | 90.53 | — | 67.27 | |
| PLOTVenue=ECCV’242024.05 | 75.28 | 90.42 | 94.12 | — | |
| OCDLImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 75.1 | 89.43 | — | 68.18 | |
| DCELImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 75.02 | 90.89 | — | — | |
| DECLVenue=MM’232024.05 | 75.02 | 90.89 | 94.52 | — | |
| FSRLVenue=ICMR’242024.05 | 74.86 | 89.97 | 94.14 | 67.57 | |
| MGRLImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 73.91 | 90.68 | — | 67.28 | |
| TBPSImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 73.54 | 88.19 | — | 65.38 | |
| IRRAImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 73.38 | 89.93 | — | 66.13 | |
| IRRAVenue=CVPR’232024.05 | 73.38 | 89.93 | 93.71 | 66.13 | |
| DM-AdapterImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 72.17 | 88.74 | — | 64.33 | |
| LuPerson-HAM + Ours (UATTA)Adaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.92 | 86.89 | 91.86 | 63.5 | |
| LuPerson-HAM + TCRAdaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.66 | 87.21 | 92.13 | 63.6 | |
| LuPerson-HAM + SARAdaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.63 | 86.87 | 91.79 | 63.4 | |
| LuPerson-HAMAdaptation Paradigm=Pure pretraining (no adaptation / finetuning)2026.04 | 70.59 | 86.89 | 91.78 | 63.39 | |
| LuPerson-HAM + SHOTAdaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.43 | 86.9 | 91.99 | 63.3 | |
| LuPerson-HAM + TentAdaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.3 | 87.02 | 91.74 | 63.26 | |
| LuPerson-HAM + CoOpAdaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.09 | 86.48 | 91.32 | 63.1 | |
| LuPerson-HAM + READAdaptation Paradigm=Pretrain-then-Adapt (Pre-Adp), Adaptation Rounds=102026.04 | 70.06 | 86.98 | 91.82 | 63.12 | |
| CFine2023.06 | 69.57 | 85.93 | 91.15 | — | |
| CFineImage Enc.=CLIP-ViT, Text Enc.=BERT2026.01 | 69.57 | 85.93 | — | — | |
| CFineVenue=TIP’232024.05 | 69.57 | 85.93 | 91.15 | — | |
| UniPTImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 68.5 | 84.67 | — | — | |
| ECCAAdaptation Paradigm=Semi-supervised Domain Adaptation2026.04 | 68.13 | 87.26 | 91.88 | — | |
| CTGIImage Enc.=CLIP-ViT, Text Enc.=CLIP-Xformer2026.01 | 67.82 | 85.45 | — | 55.14 | |
| LCR2SImage Enc.=RN50, Text Enc.=TextCNN2026.01 | 67.36 | 84.19 | — | 59.24 | |
| Baselinedescription=without pretraining on MALS2023.06 | 66.44 | 84.92 | 90.76 | 59.19 | |
| IVTPre-training=MALS (1.5M pairs)2023.06 | 66.1 | 83.79 | 89.46 | — | |
| BEATImage Enc.=RN101, Text Enc.=BERT2026.01 | 65.61 | 83.45 | — | — | |
| IVTPre-training=4M general image-text pairs2023.06 | 65.59 | 83.11 | 89.21 | — | |
| IVTImage Enc.=ViT-Base, Text Enc.=BERT2026.01 | 65.59 | 83.11 | — | 60.66 | |
| LGUR2023.06 | 65.25 | 83.12 | 89 | — | |
| SRCF2023.06 | 64.88 | 83.02 | 88.56 | — | |
| VAL-PAT2023.04 | 64.7 | 83.3 | 88.9 | — | |
| AXM-Net2023.06 | 64.44 | 80.52 | 86.77 | 58.73 | |
| CAIBC2023.06 | 64.43 | 82.87 | 88.37 | — | |
| TextReIDAdaptation Paradigm=Semi-supervised Domain Adaptation2026.04 | 64.4 | 81.27 | 87.96 | 61.19 | |
| LGUR2023.04 | 64.2 | 81.9 | 87.9 | — | |
| LBUL2023.06 | 64.04 | 82.66 | 87.22 | — | |
| SRCF2023.04 | 64 | 83 | 88.8 | — | |
| Generation-then-RetrievalAdaptation Paradigm=Semi-supervised Domain Adaptation2026.04 | 63.87 | — | — | 57.18 | |
| TIPCB2023.06 | 63.63 | 82.82 | 89.01 | — | |
| TextReIDPre-training=30k image-text pairs from MALS2023.06 | 61.81 | 80.39 | 86.9 | — | |
| TBPS2023.06 | 61.65 | 80.98 | 86.78 | — | |
| TextReIDBackbone=ResNet502023.06 | 61.65 | 80.98 | 86.78 | 58.29 | |
| NAFSconfig_version=v22023.06 | 61.5 | 81.19 | 87.51 | — | |
| SSAN2023.04 | 61.4 | 80.2 | 86.7 | — | |
| SSAN2023.06 | 61.37 | 80.15 | 86.73 | — | |
| SSANImage Enc.=RN50, Text Enc.=LSTM2026.01 | 61.37 | 80.15 | — | — | |
| NAFS*variant=*2023.04 | 61.2 | 80.6 | 87.2 | — | |
| MGEL2023.06 | 60.27 | 80.01 | 86.74 | — | |
| DSSL2023.04 | 60 | 80.4 | 87.6 | — | |
| DSSL2023.06 | 59.98 | 80.41 | 87.56 | — | |
| NAFSconfig_version=v12023.06 | 59.94 | 79.86 | 86.7 | 54.07 | |
| NAFS2023.04 | 59.9 | 79.9 | 86.7 | — | |
| MUMAAdaptation Paradigm=Unsupervised Domain Adaptation2026.04 | 59.52 | 77.79 | 84.65 | 52.75 | |
| HGAN2023.04 | 59 | 79.5 | 86.6 | — | |
| HGAN2023.06 | 59 | 79.49 | 86.62 | — | |
| SYNTH-PEDESAdaptation Paradigm=Pure pretraining (no adaptation / finetuning)2026.04 | 57.58 | — | — | 52.45 | |
| CMMTAdaptation Paradigm=Semi-supervised Domain Adaptation2026.04 | 57.1 | 78.14 | 85.23 | — | |
| CMAAM2023.04 | 56.7 | 77.2 | 84.9 | — | |
| CMAAM2023.06 | 56.68 | 77.18 | 84.86 | — | |
| IMG-Net2023.06 | 56.48 | 76.89 | 85.01 | — | |
| VITAA2023.04 | 56 | 75.8 | 83.5 | — | |
| ViTAA2023.06 | 55.97 | 75.84 | 83.52 | 51.6 | |
| IVTType=Baseline2023.06 | 55.75 | 75.68 | 84.13 | — | |
| ViTAAImage Enc.=RN50, Text Enc.=LSTM2026.01 | 54.92 | 75.18 | — | 51.6 | |
| TIMAM2023.04 | 54.5 | 77.6 | 84.8 | — | |
| PSPDAdaptation Paradigm=Unsupervised Domain Adaptation2026.04 | 53.47 | 72.81 | 76.57 | 46.41 | |
| A-GANet2023.06 | 53.14 | 74.03 | 81.95 | — | |
| MIA2023.06 | 53.1 | 75 | 82.9 | — | |
| CMPM+CMPC2023.06 | 49.37 | — | 79.21 | — | |
| GTRAdaptation Paradigm=Unsupervised Domain Adaptation2026.04 | 48.49 | 68.88 | 76.51 | 43.67 | |
| GAAPAdaptation Paradigm=Unsupervised Domain Adaptation2026.04 | 47.64 | 67.79 | 76.08 | 41.28 | |
| Dual Path2023.06 | 44.4 | 66.26 | 75.07 | — | |
| GLA2023.06 | 43.58 | 66.93 | 76.2 | — | |
| LuPerson-MLLMAdaptation Paradigm=Pure pretraining (no adaptation / finetuning)2026.04 | 38.29 | 56.6 | 64.56 | 20.43 | |
| PWM-ATH2023.06 | 27.14 | 49.45 | 61.02 | — | |
| LuPerson-TAdaptation Paradigm=Pure pretraining (no adaptation / finetuning)2026.04 | 21.88 | — | — | 19.96 | |
| GNA-RNN2023.06 | 19.05 | — | 53.64 | — | |
| CNN-RNN2023.06 | 8.07 | — | 32.47 | — | |
| CLIPAdaptation Paradigm=Pure pretraining (no adaptation / finetuning)2026.04 | 6.67 | 17.91 | — | 2.51 |