姿态引导的双分支换装行人重识别网络
|
|
周思瑶,夏楠,江佳鸿
|
Pose-guided dual-branch network for clothing-changing person re-identification
|
|
Siyao ZHOU,Nan XIA,Jiahong JIANG
|
|
| 表 1 不同方法在PRCC和VC-Clothes数据集上的性能对比 |
| Tab.1 Performance comparison of different methods on PRCC and VC-Clothes datasets |
|
| 方法 | PRCC常规场景 | | PRCC换衣场景 | | VC-Clothes常规场景 | | VC-Clothes换衣场景 | | mAP/% | Rank-1/% | | mAP/% | Rank-1/% | | mAP/% | Rank-1/% | | mAP/% | Rank-1/% | | PCB[2] | 97.0 | 99.8 | | 38.7 | 41.8 | | 74.6 | 87.4 | | 62.2 | 62.0 | | AGW[5] | 89.0 | 97.8 | | 37.1 | 39.7 | | 89.7 | 91.1 | | 82.1 | 92.0 | | CAL[12] | 99.2 | 100. 0 | | 55.8 | 55.2 | | 95.3 | 95.1 | | 87.2 | 92.9 | | TransReID[6] | 97.0 | 98.2 | | 45.0 | 42.9 | | 93.8 | 92.4 | | 81.0 | 90.4 | | CRE+BSGA[18] | 97.3 | 99.6 | | 58.7 | 61.8 | | 88.2 | 94.4 | | 84.3 | 84.5 | | SCNet[15] | 97.8 | 100. 0 | | 59.9 | 61.3 | | 89.6 | 94.9 | | 84.4 | 90.1 | | IMS-GEP[1] | 99.8 | 99.7 | | 65.8 | 57.3 | | 94.9 | 94.7 | | 81.7 | 81.8 | | CDM+GCA[7] | 94.3 | 99.3 | | 61.3 | 64.8 | | 92.8 | 93.1 | | 82.7 | 83.7 | | IRM[4] | — | — | | 52.3 | 54.2 | | — | — | | 80.1 | 90.1 | | PGAL[14] | — | — | | 58.7 | 59.5 | | — | — | | — | — | | PGNet | 99.2 | 99.8 | | 60.5 | 63.7 | | 91.3 | 95.4 | | 84.7 | 93.3 |
|
|
|