Research map / Video / Video segmentation and tracking
Semi-Supervised Video Inpainting With Cycle Consistency Constraints
Read the paperCheck 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 semi-supervised inpainting with one annotated mask rather than VOS+VI pipelines | Ours · YouTube-VOS and DAVIS | PSNR: 30.834 vs 25.255; 32.097 vs 22.696 | Helped |
| add cycle consistency loss (ablation) | Ours · YouTube-VOS | PSNR: 29.522 -> 30.834 | Helped |
| add cycle consistency loss (ablation) | OPN+MPN · YouTube-VOS | PSNR: 29.149 -> 30.017 | Helped |
| add cycle consistency loss (ablation) | CPVINet+MPN · YouTube-VOS | PSNR: 26.924 -> 27.952 | Helped |
| add cycle consistency loss (ablation) | E2FGVI+MPN · YouTube-VOS | PSNR: 28.139 -> 29.721 | Helped |
| move the known mask from the first frame to the middle or last frame (ablation) | Ours · YouTube-VOS | PSNR: 30.834; 30.826; 30.796 | No clear effect |
| add DCA feature-alignment blocks (ablation) | Ours · YouTube-VOS | PSNR: 26.667 -> 30.834 | Helped |
| increase DCA blocks from four to five or six (ablation) | Ours · YouTube-VOS | PSNR: 30.834 -> 30.839; 30.834 -> 30.851 | No clear effect |
| increase annotated frames from one to five (ablation) | Ours · YouTube-VOS | PSNR: 30.834 -> 31.086 | Helped |
| increase annotated frames from five to seven (ablation) | Ours · YouTube-VOS | PSNR: 31.086 -> 31.056 | Hurt |
Closest papers by meaning
- Deep Video Inpainting (CVPR 2019 · similarity 0.90)
- Keyframe-Guided Creative Video Inpainting (CVPR 2025 · similarity 0.84)
- Towards Language-Driven Video Inpainting via Multimodal Large Language Models (CVPR 2024 · similarity 0.84)
- CoCoCo: Improving Text-Guided Video Inpainting for Better Consistency, Controllability and Compatibility (AAAI 2025 · similarity 0.84)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.84)
- Reconstructing the Image Stitching Pipeline: Integrating Fusion and Rectangling into a Unified Inpainting Model (NeurIPS 2024 · 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.