Visible-Infrared Person Re-identification on SYSU-MM01 Indoor-search Multi-shot
88.32Rank-1 AccuracyCIFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CIFTBackbone=ResNet-50, Input Resolution=288 × 144, Inference Feature Type=transferred graph features2022.08 | 88.32 | 86.42 | |
| CIFT†Backbone=ResNet-50, Input Resolution=288 × 144, Inference Feature Type=backbone features2022.08 | 86.97 | 77.03 | |
| MPANetBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 84.22 | 75.11 | |
| SMCLBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 79.57 | 66.57 | |
| CIMABackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 73.8 | 68.3 | |
| cm-SSFTBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 73 | 72.4 | |
| HCBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 69.76 | 57.81 | |
| AlignGANBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 57.1 | 45.3 | |
| JSIA-ReIDBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 52.7 | 42.7 | |
| cmGANBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 37 | 32.76 | |
| Zero-PadBackbone=ResNet-50, Input Resolution=288 × 1442022.08 | 24.43 | 18.64 |