Research map / Label-efficient and robust learning / Self-supervised and contrastive
Pose-Disentangled Contrastive Learning for Self-Supervised Facial Representation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use pose-disentangled contrastive learning instead of SimCLR | self-supervised facial representation learning · FER-2013 | accuracy (%): 49.51 -> 56.81 | Helped |
| Use pose-disentangled contrastive learning instead of SimCLR | self-supervised facial representation learning · RAF-DB | accuracy (%): 71.06 -> 74.47 | Helped |
| Use pose-disentangled contrastive learning instead of SimCLR | self-supervised facial representation learning · LFW | accuracy (%): 75.97 -> 79.72 | Helped |
| Use pose-disentangled contrastive learning instead of SimCLR | self-supervised facial representation learning · CPLFW | accuracy (%): 62.25 -> 64.61 | Helped |
| Use pose-disentangled contrastive learning instead of SimCLR | self-supervised facial representation learning · DISFA | average F1: 49.4 -> 54.8 | Helped |
| Use pose-disentangled contrastive learning instead of SimCLR | self-supervised facial representation learning · AFLW2000 (pretrained on 300W-LP) | MAE: 14.38 -> 12.36 | Helped |
| Add pose-related contrastive loss L_pose (ablation) | PCL · RAF-DB | accuracy (%): 72.39 -> 73.73 | Helped |
| Add orthogonalizing regulation L_orth (ablation) | PCL with L_dis · RAF-DB | accuracy (%): 71.47 -> 72.39 | Helped |
| Add PDD disentanglement loss L_dis (ablation) | SimCLR baseline · RAF-DB | accuracy (%): 71.06 -> 71.47 | Helped |
| Add dynamic weighting (ablation) | PCL with L_pose · RAF-DB | accuracy (%): 73.73 -> 74.47 | Helped |
| Add pose augmentation (ablation) | SimCLR · RAF-DB | accuracy (%): 71.06 -> 73.17 | Helped |
| Add pose augmentation (ablation) | SimCLR · BU-3DFE | accuracy (%): 98.93 -> 98.85 | Hurt |
Closest papers by meaning
- Disentangled Representation Learning GAN for Pose-Invariant Face Recognition (CVPR 2017 · similarity 0.85)
- DPE: Disentanglement of Pose and Expression for General Video Portrait Editing (CVPR 2023 · similarity 0.85)
- Pose-Invariant Face Recognition via Adaptive Angular Distillation (AAAI 2022 · similarity 0.85)
- Discriminative Invariant Kernel Features: A Bells-and-Whistles-Free Approach to Unsupervised Face Recognition and Pose Estimation (CVPR 2016 · similarity 0.85)
- Self-Supervised Facial Representation Learning with Facial Region Awareness (CVPR 2024 · similarity 0.84)
- Disentangled Representation Learning via Modular Compositional Bias (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.