Research map / Video / Video segmentation and tracking
MIORe & VAR-MIORe: Benchmarks to Push the Boundaries of Restoration
George Ciubotariu, Zhuyun Zhou, Zongwei Wu, Radu Timofte
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 |
|---|---|---|---|
| increase motion blur from 1 to 249 averaged frames | AdaRevD · VAR-MIORe deblurring | PSNR: 39.61 -> 21.39 | Hurt |
| increase motion blur from 1 to 249 averaged frames | LoFormer · VAR-MIORe deblurring | PSNR: 41.41 -> 21.42 | Hurt |
| increase motion blur from 1 to 249 averaged frames | NAFNet · VAR-MIORe deblurring | PSNR: 25.69 -> 20.03 | Hurt |
| increase frame displacement from 1 to 249 | EMA-VFI · VAR-MIORe VFI | PSNR: 81.16 -> 18.91 | Hurt |
| increase frame displacement from 1 to 249 | VFIMamba · VAR-MIORe VFI | PSNR: 67.79 -> 18.91 | Hurt |
| use deblurring model at 249 frames instead of just keeping the blurry input | AdaRevD · VAR-MIORe deblurring | PSNR: 21.52 -> 21.39 | No clear effect |
| evaluate on slow (S) instead of extreme (XF) motion | VFIMamba · MIORe VFI | PSNR: 35.67 -> 30.29 | Hurt |
| evaluate on mild (S) instead of extreme (XF) motion | NAFNet · MIORe deblurring | PSNR: 26.73 -> 27.59 | Improved |
| evaluate on mild (S) instead of extreme (XF) motion | FFTformer · MIORe deblurring | PSNR: 27.14 -> 26.89 | No clear effect |
| use frequency-based FFTformer instead of NAFNet on extreme motion | FFTformer · MIORe deblurring (XF) | PSNR: 26.73 -> 27.14 | Improved |
| use MemFlow instead of VideoFlow | MemFlow · MIORe optical flow | EPE: 19.38 -> 2.99 | Improved |
| use GoPro-trained SOTA deblurring models on MIORe | FFTformer · MIORe deblurring | PSNR: 31.70 -> 27.48 | Hurt |
Closest papers by meaning
- 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)
- Posterior-Mean Rectified Flow: Towards Minimum MSE Photo-Realistic Image Restoration (ICLR 2025 · similarity 0.85)
- Turb-Seg-Res: A Segment-then-Restore Pipeline for Dynamic Videos with Atmospheric Turbulence (CVPR 2024 · similarity 0.85)
- DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework (CVPR 2025 · similarity 0.84)
- One-Step Diffusion for Detail-Rich and Temporally Consistent Video Super-Resolution (NeurIPS 2025 · similarity 0.84)
- Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models (NeurIPS 2026 · 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.
