Face Verification On Ijb A
Métriques
TAR @ FAR=0.01
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | TAR @ FAR=0.01 |
|---|---|
| face-search-at-scale-80-million-gallery | 73.30% |
| an-all-in-one-convolutional-neural-network | 92.20% |
| faceposenet-making-a-case-for-landmark-free | 90.1% |
| triplet-probabilistic-embedding-for-face | 90% |
| template-adaptation-for-face-verification-and | 93.90% |
| l2-constrained-softmax-loss-for | 97% |
| dual-agent-gans-for-photorealistic-and | 97.60% |
| inclusive-normalization-of-face-images-to | 94.60% |
| probabilistic-face-embeddings | 97.5% |
| semi-supervised-adversarial-learning-to | 53.507% |
| ghostvlad-for-set-based-face-recognition | 97.2% |
| vggface2-a-dataset-for-recognising-faces | 96.8% |
| do-we-really-need-to-collect-millions-of | 88.60% |
| unconstrained-face-verification-using-deep | 83.80% |
| face-recognition-using-deep-multi-pose | 78.70% |
| pose-robust-face-recognition-via-deep | 94.40% |
| neural-aggregation-network-for-video-face | 94.10% |