Research map / Video / Video segmentation and tracking
BVINet: Unlocking Blind Video Inpainting with Zero Annotations
Zhiliang Wu, Kerui Chen, Kun Li, Hehe Fan, Yi Yang
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 DWT down-sampling in short-term prediction module (ablation) | MPNet · mask prediction | BCE: 1.1251 -> 1.0785 | Improved |
| use DWT down-sampling in short-term prediction module (ablation) | MPNet · mask prediction | IOU: 0.8437 -> 0.8682 | Improved |
| add long-term refinement module (ablation) | MPNet DWT_STP · mask prediction | BCE: 1.0785 -> 0.9176 | Improved |
| use full MPNet (wavelet down-sampling in both modules) (ablation) | MPNet DWT_STP+LTR · mask prediction | IOU: 0.8829 -> 0.9017 | Improved |
| add sparse self-attention branch to dense branch (ablation) | VCNet DSA only · YouTube-VOS | PSNR: 29.172 -> 29.885 | Improved |
| add consistency loss L_c (ablation) | VCNet DSA+SSA · YouTube-VOS | PSNR: 29.885 -> 30.528 | Improved |
| add consistency loss L_c (ablation) | VCNet DSA+SSA · YouTube-VOS | LPIPS: 0.6891 -> 0.6556 | Improved |
| replace MPNet+FGT with BVINet | MPNet+FGT · YouTube-VOS blind | PSNR: 27.032 -> 30.528 | Improved |
| replace MPNet+WaveFormer with BVINet | MPNet+WaveFormer · DAVIS blind | PSNR: 29.794 -> 30.961 | Improved |
| use blind BVINet instead of non-blind E2FGVI | E2FGVI · YouTube-VOS | PSNR: 30.064 vs 30.528 | No clear effect |
| use VCNet instead of WaveFormer | WaveFormer · YouTube-VOS non-blind | PSNR: 33.264 -> 34.107 | Improved |
| use VCNet instead of FGT (FLOPs) | FGT · efficiency | FLOPs: 455.91G -> 396.35G | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- WaveFormer — WaveFormer: Wavelet Transformer for Noise-Robust Video Inpainting (AAAI 2024) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Semi-Supervised Video Inpainting With Cycle Consistency Constraints (CVPR 2023 · similarity 0.89)
- VCNet: A Robust Approach to Blind Image Inpainting (ECCV 2020 · similarity 0.88)
- Recurrent Temporal Aggregation Framework for Deep Video Inpainting (TPAMI 2019 · similarity 0.88)
- Internal Video Inpainting by Implicit Long-Range Propagation (ICCV 2021 · similarity 0.87)
- ZeroPatcher: Training-free Sampler for Video Inpainting and Editing (NeurIPS 2025 · similarity 0.87)
- Keyframe-Guided Creative Video Inpainting (CVPR 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.
