HyperAI초신경

홈 뉴스 최신 연구 논문 튜토리얼 데이터셋 백과사전 SOTA LLM 모델 GPU 랭킹 컨퍼런스

한국어

시스템 설정

HyperAI초신경

Face Verification On Ijb A

평가 지표

TAR @ FAR=0.01

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름	TAR @ FAR=0.01	Paper Title	Repository
Deep CNN + COTS matcher	73.30%	Face Search at Scale: 80 Million Gallery	-
All-in-one CNN	92.20%	An All-In-One Convolutional Neural Network for Face Analysis
FPN	90.1%	FacePoseNet: Making a Case for Landmark-Free Face Alignment
Triplet probabilistic embedding	90%	Triplet Probabilistic Embedding for Face Verification and Clustering
Template adaptation	93.90%	Template Adaptation for Face Verification and Identification	-
L2-constrained softmax loss	97%	L2-constrained Softmax Loss for Discriminative Face Verification
Dual-Agent GANs	97.60%	Dual-Agent GANs for Photorealistic and Identity Preserving Profile Face Synthesis	-
StyleFNM	94.60%	Inclusive normalization of face images to passport format	-
PFEfuse + match	97.5%	Probabilistic Face Embeddings
VGG + GANFaces	53.507%	Semi-supervised Adversarial Learning to Generate Photorealistic Face Images of New Identities from 3D Morphable Model
SE-GV-4-g1	97.2%	GhostVLAD for set-based face recognition
VGGFace2_ft	96.8%	VGGFace2: A dataset for recognising faces across pose and age
Synthesis as data augmentation	88.60%	Do We Really Need to Collect Millions of Faces for Effective Face Recognition?	-
DCNN	83.80%	Unconstrained Face Verification using Deep CNN Features	-
Deep multi-pose representations	78.70%	Face Recognition Using Deep Multi-Pose Representations	-
Deep Residual Equivariant Mapping	94.40%	Pose-Robust Face Recognition via Deep Residual Equivariant Mapping
NAN	94.10%	Neural Aggregation Network for Video Face Recognition	-

0 of 17 row(s) selected.