Face Search on IJB-A
94.7Rank@1All-In-One CNN
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| All-In-One CNNTPE=true2016.11 | 94.7 | 79.2 | 88.7 | — | — | |
| ResNet-18 (end2end+retrain)Training Data=full MS-Celeb-1M, Strategy=DREAM block, fine-tuned=false2018.03 | 94.6 | — | — | 96.8 | — | |
| All-In-One CNNTPE=false2016.11 | 94.1 | 70.4 | 83.6 | — | — | |
| ResNet-18 (naïve)Training Data=full MS-Celeb-1M, Strategy=naïve, fine-tuned=false2018.03 | 93.9 | — | — | 96 | — | |
| Sankaranarayanan et al.TPE=true2016.11 | 93.2 | 75.3 | 86.3 | — | — | |
| CNNmedia+TPE (f.t.)fine-tuned=true2018.03 | 93.2 | — | — | — | — | |
| Crosswhite et al.2016.11 | 92.8 | 77.4 | 88.2 | — | — | |
| Template Adaptation (f.t.)fine-tuned=true2018.03 | 92.8 | — | — | — | — | |
| Sankaranarayanan et al.TPE=false2016.11 | 92.5 | 67 | 82 | — | — | |
| ResNet-50 (end2end+retrain)Training Data=MS-Celeb-1M subset, Strategy=DREAM block, fine-tuned=false2018.03 | 92.4 | — | — | 96.2 | — | |
| ResNet-18 (end2end+retrain)Training Data=MS-Celeb-1M subset, Strategy=DREAM block, fine-tuned=false2018.03 | 91.5 | — | — | 96.2 | — | |
| VGG-Face2016.11 | 91.3 | 46 | 67 | — | — | |
| ResNet-50 (naïve)Training Data=MS-Celeb-1M subset, Strategy=naïve, fine-tuned=false2018.03 | 91.3 | — | — | 95.7 | — | |
| Augmented FacesFine-tuning=false2016.03 | 90.6 | — | — | 96.2 | 97.7 | |
| Masi et al.2016.11 | 90.6 | — | — | — | — | |
| Augmentation+Video Pooling+Rendered Testfine-tuned=false2018.03 | 90.6 | — | — | 96.2 | — | |
| Chen et al.Fine-tuning=true2016.03 | 90.3 | — | — | 96.5 | 97.7 | |
| Chen et al.2016.11 | 90.3 | — | — | — | — | |
| DCNNfusion (f.t.)fine-tuned=true2018.03 | 90.3 | — | — | 96.5 | — | |
| ResNet-18 (naïve)Training Data=MS-Celeb-1M subset, Strategy=naïve, fine-tuned=false2018.03 | 89.7 | — | — | 95.1 | — | |
| Swami S. et al.Fine-tuning=true2016.03 | 88 | — | — | 95 | — | |
| DCNN-all2016.03 | 86 | — | — | 94.3 | — | |
| DCNN-allProtocol=1:N (Search Protocol)2016.03 | 86 | — | — | 94.3 | — | |
| Multi-task CNNfine-tuned=false2018.03 | 85.8 | — | — | 93.8 | — | |
| multi-pose deep face representation2016.03 | 84.6 | 87.6 | 95.4 | 92.7 | 94.7 | |
| multi-pose deep face representationProtocol=1:N (Search Protocol)2016.03 | 84.6 | 87.6 | 95.4 | 92.7 | — | |
| Deep Multi-PoseFine-tuning=false2016.03 | 84.6 | — | — | 92.7 | 94.7 | |
| Pooling Facesfine-tuned=false2018.03 | 84.6 | — | — | 93.3 | — | |
| Deep Multi-Posefine-tuned=false2018.03 | 84.6 | — | — | 92.7 | — | |
| PAMsfine-tuned=false2018.03 | 84 | — | — | 92.5 | — | |
| Wang et al.fine-tuned=false2018.03 | 82.2 | — | — | 93.1 | — | |
| [19]2016.03 | 82 | 73.3 | 89.5 | 92.9 | — | |
| [19]Protocol=1:N (Search Protocol)2016.03 | 82 | 73.3 | 89.5 | 92.9 | 0.947 | |
| Wang et al.Fine-tuning=false2016.03 | 82 | — | — | 92.9 | — | |
| Chen et al.Fine-tuning=false2016.03 | 72.6 | — | — | 84 | 88.4 | |
| Bilinear CNNs2016.03 | 58.8 | — | — | 79.7 | — | |
| Bilinear CNNProtocol=1:N (Search Protocol)2016.03 | 58.8 | — | — | 79.7 | — | |
| COTS2016.03 | 44.3 | 40.6 | 62.7 | 59.5 | — | |
| COTSProtocol=1:N (Search Protocol)2016.03 | 44.3 | 40.6 | 62.7 | 59.5 | — | |
| GOTSFine-tuning=false2016.03 | 44.3 | — | — | 59.5 | — | |
| GOTS2016.11 | 44.3 | 4.7 | 23.5 | — | — | |
| GOTS2016.03 | 24.6 | 23.6 | 43.3 | 37.5 | — | |
| GOTSProtocol=1:N (Search Protocol)2016.03 | 24.6 | 23.6 | 43.3 | 37.5 | — | |
| OpenBRFine-tuning=false2016.03 | 24.6 | — | — | 37.5 | — | |
| multi-pose deep face representationProtocol=1:1 (Verification Protocol)2016.03 | — | 78.7 | 91.1 | — | — |