Research map / Video / Video segmentation and tracking
Semantic-Aware Dynamic Parameter for Video Inpainting Transformer
Eunhye Lee, Jinsu Yoo, Yunjeong Yang, Sungyong Baik, Tae Hyun Kim
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 super-categorized segmentation map as input (ablation) | FuseFormer-small · DAVIS | PSNR: 30.51 -> 30.90 | Improved |
| add semantic-aware dynamic linear layer naively (ablation) | FuseFormer-small · DAVIS | PSNR: 30.90 -> 30.92 | No clear effect |
| add token-wise classification loss (ablation) | SAVIT · DAVIS | PSNR: 30.92 -> 31.19 | Improved |
| add semantic-aware discriminator (ablation) | SAVIT · DAVIS | PSNR / VFID: PSNR 31.19 -> 31.01; VFID 0.165 -> 0.157 | Mixed |
| use hardmax instead of softmax in router (ablation) | SAVIT · DAVIS | PSNR: 30.91 -> 31.01 | Improved |
| replace trained routing weights with random weights (ablation) | SAVIT · DAVIS | PSNR: 31.01 -> 19.86 | Hurt |
| increase number of experts from 4 to 8 (ablation) | SAVIT · DAVIS | PSNR: 30.83 -> 31.01 | Improved |
| increase number of semantic blocks from 1 to 2 (ablation) | SAVIT · DAVIS | PSNR: 30.40 -> 30.45 | Improved |
| add semantic-aware module under large masks | FuseFormer-small · DAVIS mask ratio 60% | PSNR: 18.10 -> 19.21 | Improved |
Closest papers by meaning
- Keyframe-Guided Creative Video Inpainting (CVPR 2025 · similarity 0.88)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.87)
- Reconstruct, Inpaint, Test-Time Finetune: Dynamic Novel-view Synthesis from Monocular Videos (NeurIPS 2025 · similarity 0.87)
- Towards Language-Driven Video Inpainting via Multimodal Large Language Models (CVPR 2024 · similarity 0.87)
- CoCoCo: Improving Text-Guided Video Inpainting for Better Consistency, Controllability and Compatibility (AAAI 2025 · similarity 0.87)
- Elevating Flow-Guided Video Inpainting with Reference Generation (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.
