Person Search on PRW v1 (test)
61.2mAPSPNet-L
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SPNet-LBackbone=ConvNeXt-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=Ours-QC / COCO2024.12 | 61.2 | 90.9 | |
| SPNet-LBackbone=ConvNeXt-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=Ours-OC / COCO2024.12 | 60.7 | 90.2 | |
| SeqNetBackbone=Swin-B, Pre-training (1)=SOLIDER / LUP, Pre-training (2)=-2024.12 | 59.7 | 86.8 | |
| LEAPSBackbone=PVTv2-B2, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 59.5 | 89.7 | |
| SPNet-LBackbone=ConvNeXt-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 58.9 | 89.7 | |
| SeqNeXtBackbone=ConvNeXt-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 57.6 | 89.5 | |
| PSTRBackbone=PVTv2-B2, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 56.5 | 89.7 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=Ours-QC / COCO2024.12 | 54.2 | 89 | |
| COATBackbone=ResNet50, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 53.3 | 87.4 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=SOLIDER / LUP, Pre-training (2)=Ours-QC / COCO2024.12 | 53 | 88.3 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=Ours-OC / COCO2024.12 | 52.6 | 88.7 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 49.7 | 85.8 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=SOLIDER / COCO2024.12 | 49.7 | 86.1 | |
| SeqNetBackbone=ResNet50, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 46.7 | 83.4 | |
| SeqNetBackbone=Swin-B, Pre-training (1)=Classifier / IN1k, Pre-training (2)=-2024.12 | 45.1 | 82.5 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=SOLIDER / LUP, Pre-training (2)=-2024.12 | 38.1 | 81.3 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=SOLIDER / LUP, Pre-training (2)=SOLIDER / COCO2024.12 | 38.1 | 81.3 | |
| SPNet-LBackbone=Swin-B, Pre-training (1)=-, Pre-training (2)=-2024.12 | 20.3 | 68.7 | |
| SeqNetBackbone=Swin-B, Pre-training (1)=-, Pre-training (2)=-2024.12 | 13.8 | 55.9 |