pith. sign in

arxiv: 1708.08197 · v1 · pith:JQDKUEAPnew · submitted 2017-08-28 · 💻 cs.CV · cs.DB

Cross-Age LFW: A Database for Studying Cross-Age Face Recognition in Unconstrained Environments

classification 💻 cs.CV cs.DB
keywords facecross-agedatabaseaccuracylearningpairsverificationaging
0
0 comments X
read the original abstract

Labeled Faces in the Wild (LFW) database has been widely utilized as the benchmark of unconstrained face verification and due to big data driven machine learning methods, the performance on the database approaches nearly 100%. However, we argue that this accuracy may be too optimistic because of some limiting factors. Besides different poses, illuminations, occlusions and expressions, cross-age face is another challenge in face recognition. Different ages of the same person result in large intra-class variations and aging process is unavoidable in real world face verification. However, LFW does not pay much attention on it. Thereby we construct a Cross-Age LFW (CALFW) which deliberately searches and selects 3,000 positive face pairs with age gaps to add aging process intra-class variance. Negative pairs with same gender and race are also selected to reduce the influence of attribute difference between positive/negative pairs and achieve face verification instead of attributes classification. We evaluate several metric learning and deep learning methods on the new database. Compared to the accuracy on LFW, the accuracy drops about 10%-17% on CALFW.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 14 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Sparsity-Inducing Divergence Losses for Biometric Verification

    cs.CV 2026-06 unverdicted novelty 7.0

    Q-Margin encodes margin penalties into the reference measure of an alpha-divergence loss to produce sparse discriminative embeddings for face and speaker verification.

  2. PreFIQs: Face Image Quality Is What Survives Pruning

    cs.CV 2026-05 unverdicted novelty 7.0

    Face image quality is quantified as the Euclidean distance between embeddings from a pre-trained face recognition model and its pruned version, achieving competitive or superior results without training or supervision.

  3. FaceMoE: Mixture of Experts for Low-Resolution Face Recognition

    cs.CV 2026-06 unverdicted novelty 6.0

    FaceMoE introduces a MoE transformer with top-k routed specialized FFN experts for resolution-aware feature extraction in low-resolution face recognition, outperforming prior methods on eleven datasets.

  4. ReAge3D: Re-Aging 3D Faces with View Consistency

    cs.CV 2026-06 unverdicted novelty 6.0

    ReAge3D trains a diffusion re-aging model on synthetic pairs then uses masked propagation from a frontal pivot view to produce consistent multi-view images that supervise 3D face optimization.

  5. EX-FIQA: Leveraging Intermediate Early eXit Representations from Vision Transformers for Face Image Quality Assessment

    cs.CV 2026-04 unverdicted novelty 6.0

    Fusing quality scores from multiple intermediate transformer blocks in ViTs via depth-weighted averaging improves face image quality assessment on benchmarks without retraining or architecture changes.

  6. ATTN-FIQA: Interpretable Attention-based Face Image Quality Assessment with Vision Transformers

    cs.CV 2026-04 unverdicted novelty 6.0

    ATTN-FIQA computes face image quality scores from pre-softmax attention patterns in pre-trained ViT-based FR models using a single forward pass, showing correlation with recognition utility and spatial interpretability.

  7. Vision Transformers for Face Recognition Need More Registers

    cs.CV 2026-06 unverdicted novelty 5.0

    Adding eight register tokens to a CPE-based ViT-B for face recognition yields state-of-the-art verification accuracy on IJB-B and IJB-C while producing smoother attention maps.

  8. SteerFace: Debiasing Synthetic Face Generation via Adaptive Residue Perturbation

    cs.CV 2026-05 unverdicted novelty 5.0

    SteerFace perturbs identity embeddings toward random orthogonal directions on the hypersphere with an adaptive strategy to mitigate visual tendency in synthetic faces and improve downstream recognition performance.

  9. Reinforcement-Guided Synthetic Data Generation for Privacy-Sensitive Identity Recognition

    cs.CV 2026-04 unverdicted novelty 5.0

    A reinforcement learning approach adapts general generative models to produce synthetic data that boosts identity recognition accuracy and generalization under privacy constraints.

  10. Encoder-Decoder Manifold Alignment for Idempotent Generation

    cs.LG 2026-06 unverdicted novelty 4.0

    Encoder-decoder manifold alignment framework to achieve exact idempotency in generative models.

  11. ViT-FREE: Efficient Face Recognition via Early Exiting and Synthetic Adaptation

    cs.CV 2026-06 conditional novelty 4.0

    ViT-FREE enables early exiting from pretrained ViTs for face verification with up to 20% speedup and 1.5 accuracy drop on IJB-C, plus a synthetic-data fine-tuning variant for shallow exits.

  12. Lightweight Cross-Spectral Face Recognition via Contrastive Alignment and Distillation

    cs.CV 2026-05 unverdicted novelty 4.0

    A lightweight hybrid CNN-Transformer framework for heterogeneous face recognition achieves competitive performance on cross-spectral benchmarks and standard RGB tasks using contrastive alignment and distillation.

  13. FunFace: Feature Utility and Norm Estimation for Face Recognition

    cs.CV 2026-04 unverdicted novelty 4.0

    FunFace is a new loss function for face recognition training that incorporates biometric utility via Certainty Ratio into adaptive margins to improve results on low-quality samples.

  14. FaceLiVTv2: An Improved Hybrid Architecture for Efficient Mobile Face Recognition

    cs.CV 2026-04 unverdicted novelty 4.0

    FaceLiVTv2 improves the accuracy-efficiency trade-off for mobile face recognition, cutting inference latency by 22% versus its predecessor while outperforming other lightweight models on standard benchmarks.