DIME: Scaling Facial Representation Learning via Differential Masked Autoencoding
cs.CV
Submitted: 2026-10-12
Updated: 2026-10-12
Project page: https://cv-ac.github.io/DIME
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction
- DINOv3
- DINOv2: Learning Robust Visual Features without Supervision
- Representation Learning and Identity Adversarial Training for Facial Behavior Understanding
- PaCo-FR: Patch-Pixel Aligned End-to-End Codebook Learning for Facial Representation Pre-training
- FaceMAE: Privacy-Preserving Face Recognition via Masked Autoencoders
- A Generative Framework for Self-Supervised Facial Representation Learning
- GPTFace: Generative Pre-training of Facial-Linguistic Transformer by Span Masking and Weakly Correlated Text-image Data
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Learning Multi-dimensional Edge Feature-based AU Relation Graph for Facial Action Unit Recognition
- RankPose: Learning Generalised Feature with Rank Supervision for Head Pose Estimation
- Improved Baselines with Representation Autoencoders
- Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think
- Scaling Laws for Neural Language Models
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models