Research map / 3D vision / 3D generation
3D-Aware Facial Landmark Detection via Multi-View Consistent Training on Synthetic Data
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 |
|---|---|---|---|
| Add 3D-aware multi-view consistency training to DAD-3DNet | DAD-3DNet · DAD-3DHeads | NME: 2.599 -> 2.503 | Helped |
| Add 3D-aware multi-view consistency training to DAD-3DNet | DAD-3DNet · FaceScape | NME: 6.681 -> 6.050 | Helped |
| Add 3D-aware multi-view consistency training to DAD-3DNet | DAD-3DNet · MultiFace | NME: 5.786 -> 5.480 | Helped |
| Add 3D-aware multi-view consistency training to DAD-3DNet | DAD-3DNet · DAD-3DHeads | Overall head pose error: 7.382 -> 6.500 | Helped |
| Add 3D-aware multi-view consistency training to DAD-3DNet | DAD-3DNet · FaceScape | Overall head pose error: 14.624 -> 11.863 | Helped |
| Add mesh consistency loss (ablation) | 3D-aware DAD-3DNet training · FaceScape | NME: 6.168 -> 6.050 | Helped |
| Add mesh consistency loss (ablation) | 3D-aware DAD-3DNet training · FaceScape | Pose: 12.327 -> 11.863 | Helped |
| Add self-projection consistency loss (ablation) | 3D-aware DAD-3DNet training · FaceScape | NME: 6.541 -> 6.050 | Helped |
| Add self-projection consistency loss (ablation) | 3D-aware DAD-3DNet training · FaceScape | Pose: 13.623 -> 11.863 | Helped |
| Use P=8 views rather than P=4 to estimate 3D landmarks (ablation) | 3D-aware DAD-3Net training · FaceScape | NME: 6.050 -> 6.048 | No clear effect |
| Use P=8 views rather than P=4 to estimate 3D landmarks (ablation) | 3D-aware DAD-3Net training · FaceScape | Pose: 11.863 -> 11.923 | No clear effect |
| Use P=16 views rather than P=4 to estimate 3D landmarks (ablation) | 3D-aware DAD-3Net training · FaceScape | NME: 6.050 -> 6.098 | Hurt |
Closest papers by meaning
- Continuous Landmark Detection With 3D Queries (CVPR 2023 · similarity 0.90)
- Learning to Detect 3D Facial Landmarks via Heatmap Regression with Graph Convolutional Network (AAAI 2022 · similarity 0.88)
- Brute-Force Facial Landmark Analysis With a 140,000-Way Classifier (AAAI 2018 · similarity 0.87)
- FaceLift: Semi-supervised 3D Facial Landmark Localization (CVPR 2024 · similarity 0.87)
- Semantic Alignment: Finding Semantically Consistent Ground-Truth for Facial Landmark Detection (CVPR 2019 · similarity 0.87)
- T-FAKE: Synthesizing Thermal Images for Facial Landmarking (CVPR 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.