Research map / Video / Video segmentation and tracking
Occlusion-Aware Video Object Inpainting
Lei Ke, Yu-Wing Tai, Chi-Keung Tang
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 occlusion-aware gating (ablation) | VOIN baseline · YouTube-VOI (VOIN component ablation) | PSNR: 44.75 -> 45.25 | Improved |
| add occlusion-aware gating (ablation) | VOIN baseline · YouTube-VOI (VOIN component ablation) | LPIPS: 0.025 -> 0.019 | Improved |
| add T-PatchGAN discriminator (ablation) | VOIN with gating · YouTube-VOI (VOIN component ablation) | PSNR: 45.25 -> 45.54 | Improved |
| add multi-class discriminator (ablation) | VOIN with gating and T-PatchGAN · YouTube-VOI (VOIN component ablation) | PSNR: 45.54 -> 45.91 | Improved |
| add spatio-temporal attention module (ablation) | VOIN with multi-class discriminator · YouTube-VOI (VOIN component ablation) | PSNR: 45.91 -> 46.33 | Improved |
| add flow-guided pixel propagation (ablation) | VOIN full · YouTube-VOI (VOIN component ablation) | PSNR: 46.33 -> 48.99 | Improved |
| add flow-guided propagation (ablation) | VOIN full · YouTube-VOI (VOIN component ablation) | LPIPS: 0.013 -> 0.008 | Improved |
| increase transformer layers from 4 to 8 (ablation) | shape completion module · YouTube-VOI (shape completion ablation) | mIoU (%): 78.02 -> 80.28 | Improved |
| add dice loss (ablation) | shape completion module · YouTube-VOI (shape completion ablation) | mIoU (%): 80.28 -> 81.53 | Improved |
| add semantic guidance (ablation) | shape completion module · YouTube-VOI (shape completion ablation) | mIoU (%): 81.53 -> 86.68 | Improved |
| use video shape completion instead of image-based UNet (ablation) | shape completion · YouTube-VOI (shape completion ablation) | mIoU (%): 74.81 -> 86.68 | Improved |
| add amodal shape guidance (ablation) | flow completion module · YouTube-VOI (flow completion ablation) | EPE: 4.89 -> 3.95 | Improved |
Closest papers by meaning
- Proposal-based Video Completion (ECCV 2020 · similarity 0.89)
- Infinite-Canvas: Higher-Resolution Video Outpainting with Extensive Content Generation (AAAI 2025 · similarity 0.85)
- Unboxed: Geometrically and Temporally Consistent Video Outpainting (CVPR 2025 · similarity 0.85)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.85)
- Elevating Flow-Guided Video Inpainting with Reference Generation (AAAI 2025 · similarity 0.85)
- Keyframe-Guided Creative Video Inpainting (CVPR 2025 · similarity 0.85)
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.
