Research map / Vision-language and multimodal / Cross-modal learning
DADM: Dual Alignment of Domain and Modality for Face Anti-spoofing
Jingyi Yang, Xun Lin, Zitong Yu, Liepiao Zhang, Xin Liu, Hui Li, Xiaochen Yuan, Xiaochun Cao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add U-Adapter and SSP loss to vanilla ViT (ablation) | ViT · Protocol 1 average | HTER (%): 31.14 -> 24.54 | Improved |
| add Mutual Information Mask with CDC-Adapter instead of U-Adapter (ablation) | ViT + CE+SSP · Protocol 1 average | HTER (%): 24.54 -> 21.75 | Improved |
| use CDC adapter instead of vanilla conv adapter in MIM (ablation) | DADM · Protocol 1 average | HTER (%): 14.71 -> 13.63 | Improved |
| add MI-guided ReGrad and MI loss (ablation) | ViT + MIM + CDC-Adapter · Protocol 1 average | HTER (%): 21.75 -> 17.17 | Improved |
| replace CE+SSP with PG-IRM optimization (ablation) | DADM · Protocol 1 average | HTER (%): 17.17 -> 16.54 | Improved |
| add dual alignment of hyperplane and angle margin (ablation) | DADM · Protocol 1 average | HTER (%): 16.54 -> 14.31 | Improved |
| add MI loss on top of dual alignment (ablation) | DADM · Protocol 1 average | HTER (%): 14.31 -> 13.63 | Improved |
| use proposed MI loss instead of InfoNCE (ablation) | DADM · Protocol 1 average | HTER (%): 15.80 -> 13.63 | Improved |
| use proposed MI loss instead of MINE (ablation) | DADM · Protocol 1 average | HTER (%): 14.40 -> 13.63 | Improved |
| use mutual-information mask instead of threshold-based mask (ablation) | DADM · Protocol 1 average | HTER (%): 21.28 -> 13.63 | Improved |
| use mutual-information mask instead of attention-based mask (ablation) | DADM · Protocol 1 average | HTER (%): 15.74 -> 13.63 | Improved |
| use mutual-information mask instead of distance-based mask (ablation) | DADM · Protocol 1 average | HTER (%): 14.08 -> 13.63 | Improved |
Later papers that compared with this method
1 later accepted paper reports DADM as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- PA-FAS: Towards Interpretable and Generalizable Multimodal Face Anti-Spoofing via Path-Augmented Reinforcement Learning (AAAI 2026 · better on 2 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Purify then Guide: Rethinking Domain Generalization for Multimodal Face Anti-Spoofing (ECCV 2026 · similarity 0.91)
- Reliable and Balanced Transfer Learning for Generalized Multimodal Face Anti-Spoofing (TPAMI 2025 · similarity 0.90)
- Gradient Alignment for Cross-Domain Face Anti-Spoofing (CVPR 2024 · similarity 0.89)
- mmFAS: Multimodal Face Anti-Spoofing Using Multi-Level Alignment and Switch-Attention Fusion (AAAI 2025 · similarity 0.89)
- Rethinking Generalizable Face Anti-spoofing via Hierarchical Prototype-guided Distribution Refinement in Hyperbolic Space (CVPR 2024 · similarity 0.87)
- Test-Time Domain Generalization for Face Anti-Spoofing (CVPR 2024 · similarity 0.87)
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.
