Research map / 3D vision / Human pose estimation
Visibility Constrained Generative Model for Depth-Based 3D Facial Pose Tracking
Lu Sheng, Jianfei Cai, Tat‐Jen Cham, Vladimir Pavlović, King Ngi Ngan
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 |
|---|---|---|---|
| use the proposed ray-visibility tracker instead of Sheng et al. | depth-based face pose tracking · Biwi | translation error (mm): 6.9 -> 5.5 | Improved |
| use proposed tracker instead of Meyer et al. | depth-based face pose tracking · Biwi | yaw error (deg): 2.1 -> 1.6 | Improved |
| use proposed depth-only tracker instead of RGB+depth Li et al. | depth-based face pose tracking · ICT-3DHP | yaw error (deg): 3.3 -> 2.5 | Improved |
| use proposed tracker instead of Li et al. | depth-based face pose tracking · Biwi | roll error (deg): 3.2 -> 1.8 | Improved |
| use proposed tracker instead of Li et al. on pitch | depth-based face pose tracking · Biwi | pitch error (deg): 1.7 -> 1.7 | No clear effect |
| add temporal coherence and scale accumulation to Sheng et al.'s tracker (ablation) | RVS tracker · Biwi | translation accuracy: 6.9 -> 5.5 | Improved |
| add personalized (online-adapted) face model instead of generic model (ablation) | RVS tracker · Biwi and ICT-3DHP | angle and translation error | Improved |
| use ray visibility score instead of ICP (ablation) | face pose estimation · unconstrained poses with occlusion | pose quality | Improved |
| add temporal coherence term to RVS (RVS+TE) (ablation) | RVS tracker · runtime | time per frame (sec): 0.0193 -> 0.0233 | Mixed |
| add PSO to RVS+TE (ablation) | RVS tracker · runtime | time per frame (sec): 0.0233 -> 0.328 | Hurt |
| replace ICP with RVS | pose estimator · runtime | time per frame (sec): 0.0156 -> 0.0193 | No clear effect |
| add online identity adaptation | tracker · runtime | time per frame (sec): 0.0233 -> 0.131 | Hurt |
Closest papers by meaning
- A Generative Model for Depth-Based Robust 3D Facial Pose Tracking (CVPR 2017 · similarity 0.97)
- HeadFusion: 360 Head Pose Tracking Combining 3D Morphable Model and 3D Reconstruction (TPAMI 2018 · similarity 0.86)
- Unsupervised Depth Estimation, 3D Face Rotation and Replacement (NeurIPS 2018 · similarity 0.86)
- Face Anything: 4D Face Reconstruction from Any Image Sequence (ECCV 2026 · similarity 0.86)
- 3D Face Tracking from 2D Video through Iterative Dense UV to Image Flow (CVPR 2024 · similarity 0.85)
- DiffPortrait3D: Controllable Diffusion for Zero-Shot Portrait View Synthesis (CVPR 2024 · 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.
