Research map / Label-efficient and robust learning / Recognition and long-tail
Face Anti-Spoofing with Human Material Perception
Zitong Yu, Xiaobai Li, Xuesong Niu, Jingang Shi, Guoying Zhao
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 reflection-head supervision to the depth head (ablation) | Baseline D · OULU-NPU Protocol 1 | ACER (%): 4.1 -> 2.8 | Improved |
| add patch-head supervision to the depth head (ablation) | Baseline D · OULU-NPU Protocol 1 | ACER (%): 4.1 -> 2.3 | Improved |
| add reflection head to depth+patch heads (ablation) | D+P · OULU-NPU Protocol 1 | ACER (%): 2.3 -> 1.8 | Improved |
| add multi-level feature refinement module (MFRM) (ablation) | D+R+P · OULU-NPU Protocol 1 | ACER (%): 1.8 -> 1.2 | Improved |
| add bilateral convolutional networks (BCN) (ablation) | D+R+P+MFRM · OULU-NPU Protocol 1 | ACER (%): 1.2 -> 0.8 | Improved |
| use spatial attention instead of context-aware reassembling in MFRM (ablation) | D+R+P · OULU-NPU Protocol 1 | ACER (%): 1.8 -> 1.5 | Improved |
| use channel attention as the refinement method (ablation) | D+R+P · OULU-NPU Protocol 1 | ACER (%): 1.8 -> 9.5 | Hurt |
| use non-local attention as the refinement method (ablation) | D+R+P · OULU-NPU Protocol 1 | ACER (%): 1.8 -> 12.7 | Hurt |
| add bilateral operator to the extra parameter blocks of BCN (ablation) | BCN · OULU-NPU Protocol 1 | ACER (%): 4.3 -> 2.1 | Improved |
| add bilateral residual branch to BCN (ablation) | BCN · OULU-NPU Protocol 1 | ACER (%): 4.0 -> 2.1 | Improved |
| add spatial neighborhood distance term to the deep bilateral operator (ablation) | BCN · OULU-NPU Protocol 1 | ACER (%): 2.4 -> 2.1 | No clear effect |
| use bilateral residual images as inputs instead of RGB (ablation) | RGB baseline · OULU-NPU Protocol 1 | ACER (%): 0.4% ACER lower | Improved |
Closest papers by meaning
- Deep Learning for Face Anti-Spoofing: A Survey (TPAMI 2022 · similarity 0.89)
- mmFAS: Multimodal Face Anti-Spoofing Using Multi-Level Alignment and Switch-Attention Fusion (AAAI 2025 · similarity 0.87)
- FaceShield: Explainable Face Anti-Spoofing with Multimodal Large Language Models (AAAI 2026 · similarity 0.87)
- One-Class Face Anti-spoofing via Spoof Cue Map-Guided Feature Learning (CVPR 2024 · similarity 0.86)
- SLIP: Spoof-Aware One-Class Face Anti-Spoofing with Language Image Pretraining (AAAI 2025 · similarity 0.85)
- Interpretable Face Anti-Spoofing: Enhancing Generalization with Multimodal Large Language Models (AAAI 2025 · similarity 0.85)
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.
