Research map / Video / Video segmentation and tracking
Deep Flow-Guided Video Inpainting
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 |
|---|---|---|---|
| use flow completion followed by flow-guided pixel propagation instead of direct frame-wise image inpainting | Video inpainting · YouTube-VOS fixed-region test | PSNR: 16.68 -> 27.49 | Helped |
| use flow completion followed by flow-guided pixel propagation instead of direct frame-wise image inpainting | Video inpainting · DAVIS fixed-region test | PSNR: 16.47 -> 28.26 | Helped |
| add hard flow example mining with p=50% (ablation) | DFC-S · YouTube-VOS validation fixed-region inpainting | PSNR: 24.43 -> 26.15 | Helped |
| add hard flow example mining with p=50% (ablation) | DFC-S · YouTube-VOS validation flow completion | EPE (overall): 1.03 -> 0.99 | Helped |
| increase the number of stacked refinement stages from one to two (ablation) | DFC-Net · YouTube-VOS validation fixed-region inpainting | PSNR: 26.15 -> 27.10 | Helped |
| add the third coarse-to-fine refinement stage (ablation) | DFC-Net · YouTube-VOS validation flow completion | EPE: 0.94 -> 0.93 | Helped |
| replace a single ResNet-101 stage with stacked refinement stages (ablation) | DFC-Net · YouTube-VOS validation fixed-region inpainting | PSNR: 26.58 -> 27.50 | Helped |
| use multi-scale inputs across refinement stages (ablation) | DFC-Net · YouTube-VOS validation fixed-region inpainting | PSNR: 27.02 -> 27.50 | Helped |
| add flow-guided pixel propagation before hallucinating unseen regions (ablation) | Video inpainting · YouTube-VOS validation fixed-region inpainting | PSNR: 19.43 -> 27.50 | Helped |
| use multiple input frames in the Deepfill architecture (ablation) | Deepfill · YouTube-VOS fixed-region inpainting | PSNR: 16.68 -> 16.71 | No clear effect |
| stack three multi-frame Deepfill networks (ablation) | Deepfill+Multi-Frame · YouTube-VOS fixed-region inpainting | PSNR: 16.71 -> 17.02 | Helped |
| use FlowNet2 for initial flow estimation (ablation) | Huang et al. · DAVIS fixed-region test | PSNR: 27.39 -> 27.73 | Helped |
Closest papers by meaning
- Elevating Flow-Guided Video Inpainting with Reference Generation (AAAI 2025 · similarity 0.90)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.90)
- Keyframe-Guided Creative Video Inpainting (CVPR 2025 · similarity 0.88)
- HomoGen: Enhanced Video Inpainting via Homography Propagation and Diffusion (CVPR 2025 · similarity 0.87)
- Infinite-Canvas: Higher-Resolution Video Outpainting with Extensive Content Generation (AAAI 2025 · similarity 0.86)
- CoCoCo: Improving Text-Guided Video Inpainting for Better Consistency, Controllability and Compatibility (AAAI 2025 · similarity 0.86)
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.