Research map / 3D vision / Human pose estimation
Neural Head Reenactment with Latent Pose Descriptors
Check 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 |
|---|---|---|---|
| Learn 256-dimensional latent pose descriptors using reconstruction losses | Pose-based retrieval · Multi-PIE | Top-10 retrieval accuracy (%): 61.0 -> 75.7 | Helped |
| Learn 256-dimensional latent pose descriptors using reconstruction losses | Pose-based retrieval · Multi-PIE | Top-20 retrieval accuracy (%): 55.7 -> 63.8 | Helped |
| Learn 256-dimensional latent pose descriptors using reconstruction losses | Pose-based retrieval · Multi-PIE | Top-50 retrieval accuracy (%): 51.8 -> 57.8 | Helped |
| Learn 256-dimensional latent pose descriptors using reconstruction losses | Pose-based retrieval · Multi-PIE | Top-100 retrieval accuracy (%): 49.4 -> 54.1 | Helped |
| Reduce pose-vector dimensionality from 256 to 64 (ablation) | Ours · VoxCeleb2 test cross-person reenactment | pose reconstruction error | Hurt |
| Increase pose encoder capacity to ResNeXt-50 (32 × 4) (ablation) | Ours · VoxCeleb2 test cross-person reenactment | identity error | Hurt |
| Increase pose encoder capacity to ResNeXt-50 (32 × 4) (ablation) | Ours · VoxCeleb2 test cross-person reenactment | pose reconstruction error | Helped |
| Remove foreground segmentation and retain background in reconstruction (ablation) | Ours · VoxCeleb2 test cross-person reenactment | identity error | Hurt |
| Remove random pose augmentations (ablation) | Ours · VoxCeleb2 test cross-person reenactment | identity error | Hurt |
| Remove random pose augmentations (ablation) | Ours · VoxCeleb2 test cross-person reenactment | pose reconstruction error | Helped |
| Remove random pose augmentations (ablation) | X2Face+ · VoxCeleb2 test cross-person reenactment | identity error | Hurt |
| Remove random pose augmentations (ablation) | X2Face+ · VoxCeleb2 test cross-person reenactment | pose reconstruction error | Helped |
Closest papers by meaning
- FSRT: Facial Scene Representation Transformer for Face Reenactment from Factorized Appearance Head-pose and Facial Expression Features (CVPR 2024 · similarity 0.86)
- FDN: Feature Decoupling Network for Head Pose Estimation (AAAI 2020 · similarity 0.85)
- Pose-Dependent Low-Rank Embedding for Head Pose Estimation (AAAI 2016 · similarity 0.85)
- VOODOO 3D: Volumetric Portrait Disentanglement For One-Shot 3D Head Reenactment (CVPR 2024 · similarity 0.85)
- Neural Head Avatars From Monocular RGB Videos (CVPR 2022 · similarity 0.84)
- Neural Reprojection Error: Merging Feature Learning and Camera Pose Estimation (CVPR 2021 · 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.