Research map / Video / Video segmentation and tracking
Recurrent Video Restoration Transformer with Guided Deformable Attention
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 |
|---|---|---|---|
| increase clip length from 1 to 2 (ablation) | RVRT · REDS (ablation) | PSNR (dB): 31.98 -> 32.10 | Helped |
| increase clip length from 2 to 3 (ablation) | RVRT · REDS (ablation) | PSNR (dB): 32.10 -> 32.07 | No clear effect |
| replace optical-flow warping with GDA for alignment (ablation) | RVRT · REDS (ablation) | PSNR (dB): 28.88 -> 32.10 | Helped |
| replace deformable convolution with GDA (ablation) | RVRT · REDS (ablation) | PSNR (dB): 31.93 -> 32.10 | Helped |
| align from whole clip instead of frame by frame in GDA (ablation) | GDA · REDS (ablation) | PSNR (dB): 32.00 -> 32.10 | Helped |
| add optical flow guidance to GDA (ablation) | GDA · REDS (ablation) | PSNR (dB): 30.99 -> 32.03 | Helped |
| add optical flow update across layers (ablation) | GDA · REDS (ablation) | PSNR (dB): 32.03 -> 32.10 | Helped |
| add channel-interaction MLP in GDA (ablation) | GDA · REDS (ablation) | PSNR (dB): 31.83 -> 32.10 | Helped |
| increase deformable groups from 1 to 12 (ablation) | GDA · REDS (ablation) | PSNR (dB): 31.63 -> 32.10 | Helped |
| increase attention heads from 12 to 36 (ablation) | GDA with 12 groups · REDS (ablation) | PSNR (dB): 32.10 -> 32.03 | Hurt |
| RVRT instead of VRT | video SR · REDS4 x4 | PSNR (dB): 32.19 -> 32.75 | Helped |
| RVRT instead of VRT | video denoising · Set8 sigma=10 | PSNR (dB): 37.88 -> 37.53 | Mixed |
Closest papers by meaning
- Learning Truncated Causal History Model for Video Restoration (NeurIPS 2024 · similarity 0.88)
- SeedVR: Seeding Infinity in Diffusion Transformer Towards Generic Video Restoration (CVPR 2025 · similarity 0.85)
- Blur-aware Spatio-temporal Sparse Transformer for Video Deblurring (CVPR 2024 · similarity 0.85)
- FMA-Net: Flow-Guided Dynamic Filtering and Iterative Feature Refinement with Multi-Attention for Joint Video Super-Resolution and Deblurring (CVPR 2024 · similarity 0.85)
- AverNet: All-in-one Video Restoration for Time-varying Unknown Degradations (NeurIPS 2024 · similarity 0.84)
- AsymRnR: Video Diffusion Transformers Acceleration with Asymmetric Reduction and Restoration (ICML 2025 · 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.