Research map / 3D vision / Face and shape
DaGAN++: Depth-Aware Generative Adversarial Network for Talking Head Video Generation
Fa-Ting Hong, Li Shen, Dan Xu
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 improved face depth network FDN++ with uncertainty (ablation) | Baseline · HDTF ablation | SSIM (%): 81.0 -> 84.3 | Improved |
| add original DaGAN depth network FDN (ablation) | Baseline · HDTF ablation | SSIM (%): 81.0 -> 81.3 | Improved |
| add single-layer cross-attention CAM (ablation) | Baseline · HDTF ablation | SSIM (%): 81.0 -> 82.0 | Improved |
| add CAM++ with improved depth map (ablation) | Baseline · HDTF ablation | SSIM (%): 81.0 -> 83.2 | Improved |
| add multi-layer geometry-enhanced generation GMG (ablation) | Baseline · HDTF ablation | SSIM (%): 81.0 -> 86.0 | Improved |
| combine FDN++ keypoints and GMG (ablation) | Baseline w/ GMG · HDTF ablation | SSIM (%): 86.0 -> 86.7 | Improved |
| replace DaGAN with DaGAN++ | DaGAN · HDTF self-reenactment | SSIM (%): 82.3 -> 86.7 | Improved |
| replace DaGAN with DaGAN++ | DaGAN · VoxCeleb1 self-reenactment | SSIM (%): 80.4 -> 82.6 | Improved |
| replace DaGAN with DaGAN++ | DaGAN · VoxCeleb2 self-reenactment | AKD: 1.865 -> 1.675 | Improved |
| use 20 keypoints instead of 15 (ablation) | DaGAN++ · HDTF keypoint number | SSIM (%): 86.7 -> 86.6 | No clear effect |
| use 10 keypoints instead of 15 (ablation) | DaGAN++ · HDTF keypoint number | SSIM (%): 86.7 -> 86.3 | Hurt |
| improved depth gives more stable consecutive-frame depth | DaGAN · VoxCeleb1 same-id reenactment | TCM: 0.1631 -> 0.1806 | Improved |
Closest papers by meaning
- Depth-Aware Generative Adversarial Network for Talking Head Video Generation (CVPR 2022 · similarity 0.97)
- FACIAL: Synthesizing Dynamic Talking Face With Implicit Attribute Learning (ICCV 2021 · similarity 0.90)
- Hierarchical Cross-Modal Talking Face Generation With Dynamic Pixel-Wise Loss (CVPR 2019 · similarity 0.90)
- High-Fidelity and Freely Controllable Talking Head Video Generation (CVPR 2023 · similarity 0.89)
- AnyTalk: Multi-modal Driven Multi-domain Talking Head Generation (AAAI 2025 · similarity 0.87)
- IM-Portrait: Learning 3D-aware Video Diffusion for Photorealistic Talking Heads from Monocular VideosC (CVPR 2025 · 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.
