Research map / Video / Video generation
Internal Video Inpainting by Implicit Long-Range Propagation
Hao Ouyang, Tengfei Wang, Qifeng Chen
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 coarse-to-fine network instead of base network (ablation) | internal video inpainting · DAVIS fixed masks (complex sequences) | PSNR: 25.46 -> 28.54 | Improved |
| use coarse-to-fine network instead of base network (ablation) | internal video inpainting · DAVIS fixed masks (simple sequences) | PSNR: 30.05 -> 32.70 | Improved |
| stack the base network twice (doubling parameters) (ablation) | base network · DAVIS fixed masks (complex sequences) | PSNR: 25.46 -> 27.27 | Improved |
| add ambiguity loss to reconstruction loss (ablation) | Ours · DAVIS | PSNR: 29.36 -> 29.21 | Hurt |
| add stabilization loss to reconstruction loss (ablation) | Ours · DAVIS | PSNR: 29.36 -> 29.08 | Hurt |
| add ambiguity loss to reconstruction loss (ablation) | Ours · DAVIS | SSIM: 0.932 -> 0.931 | No clear effect |
| add stabilization loss to reconstruction loss (ablation) | Ours · DAVIS | SSIM: 0.932 -> 0.929 | Hurt |
| replace augmented object masks with random free-form masks in training (ablation) | Ours · DAVIS | inpainting quality | Hurt |
| use Ours instead of FGVC on object masks | DAVIS object masks · DAVIS (object masks) | PSNR: 33.98 -> 31.09 | Hurt |
| use Ours instead of InterVI | internal learning · DAVIS (fixed masks) | PSNR: 26.89 -> 29.90 | Improved |
| increase video length from 5 to 80 frames (ablation) | over-fitting ability of base network · BMX-TREES | over-fitting PSNR: 48.78 -> 41.20 | Hurt |
Closest papers by meaning
- An Internal Learning Approach to Video Inpainting (ICCV 2019 · similarity 0.92)
- Deep Video Inpainting (CVPR 2019 · similarity 0.90)
- A Temporally-Aware Interpolation Network for Video Frame Inpainting (TPAMI 2019 · similarity 0.90)
- Video Inpainting by Jointly Learning Temporal Structure and Spatial Details (AAAI 2019 · similarity 0.89)
- Learning Joint Spatial-Temporal Transformations for Video Inpainting (ECCV 2020 · similarity 0.89)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.88)
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.
