Research map / Video / Video generation
Multi-Modality Deep Restoration of Extremely Compressed Face Videos
Xi Zhang, Xiaolin Wu
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 motion vectors for frame alignment (ablation) | MDVD-Net · Obama | PSNR (dB): 0.08dB | Improved |
| add motion vectors for frame alignment (ablation) | MDVD-Net · VoxCeleb2 | PSNR (dB): 0.03dB | Improved |
| add speech processing branch (ablation) | MDVD-Net · Obama | PSNR (dB): 0.27dB | Improved |
| add speech processing branch (ablation) | MDVD-Net · VoxCeleb2 | PSNR (dB): 0.16dB | Improved |
| add facial landmark correction by speech (vs uncorrected landmarks) (ablation) | MDVD-Net · Obama | PSNR (dB): 30.31 -> 30.43 | Improved |
| add facial landmark correction by speech (vs uncorrected landmarks) (ablation) | MDVD-Net · VoxCeleb2 | PSNR (dB): 27.04 -> 27.12 | Improved |
| add spatial attention fusion (ablation) | MDVD-Net · Obama | PSNR (dB): 0.11dB | Improved |
| add spatial attention fusion (ablation) | MDVD-Net · VoxCeleb2 | PSNR (dB): 0.08dB | Improved |
| add video codec information (ablation) | MDVD-Net · Obama | PSNR (dB): 0.18dB | Improved |
| add video codec information (ablation) | MDVD-Net · VoxCeleb2 | PSNR (dB): 0.11dB | Improved |
| add speech prior, effect on lower vs upper face (ablation) | MDVD-Net · Obama (SPB) | PSNR (dB): 0.38 vs 0.16 | Improved |
| lower bitrate (60kbps vs 120kbps) increases MDVD-Net gain over EDVR | MDVD-Net · Obama | PSNR gain over EDVR (dB): 0.7dB at 60kbps -> about 0.1dB at 120kbps | Improved |
Closest papers by meaning
- DAVD-Net: Deep Audio-Aided Video Decompression of Talking Heads (CVPR 2020 · similarity 0.92)
- Audio-Assisted Face Video Restoration with Temporal and Identity Complementary Learning (AAAI 2026 · similarity 0.88)
- Efficient Video Face Enhancement with Enhanced Spatial-Temporal Consistency (CVPR 2025 · similarity 0.86)
- Dynamic Content Prediction with Motion-aware Priors for Blind Face Video Restoration (CVPR 2025 · similarity 0.85)
- Discrete Prior-Based Temporal-Coherent Content Prediction for Blind Face Video Restoration (AAAI 2025 · similarity 0.84)
- From Faces to Voices: Learning Hierarchical Representations for High-quality Video-to-Speech (CVPR 2025 · 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.
