Research map / Video / Video segmentation and tracking
Blind Video Super-Resolution based on Implicit Kernels
Qiang Zhu, Yuxuan Jiang, Shuyuan Zhu, Fan Zhang, David Bull, Bing Zeng
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 Implicit Temporal Alignment module (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 28.86 -> 29.48 | Improved |
| add Implicit Spatial Correction module (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.18 -> 29.48 | Improved |
| add recurrent hidden-state structure in the Transformer (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.20 -> 29.48 | Improved |
| add bidirectional propagation (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.29 -> 29.48 | Improved |
| add correction loss L_C (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.40 -> 29.48 | Improved |
| use INR multi-scale kernel dictionary instead of DKP kernel prediction (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 28.92 -> 29.48 | Improved |
| use ITA instead of FGDA alignment (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.45 -> 29.48 | No clear effect |
| increase kernel scales R beyond 8 (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.44 (R=8) -> 29.42 (R=9) | Hurt |
| increase atom number N from 8 to 12 (ablation) | BVSR-IK · REDS4 realistic motion blur | PSNR (dB): 29.48 -> 29.53 | Improved |
| BVSR-IK vs FMA-Net under noise sigma=25 | BVSR-IK · REDS4 with noise | PSNR (dB): 23.97 -> 24.96 | Improved |
| BVSR-IK vs FMA-Net on UDM10 Gaussian blur | BVSR-IK · UDM10 Gaussian blur | PSNR (dB): 30.19 -> 30.21 | No clear effect |
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.
- FMA-Net — FMA-Net: Flow-Guided Dynamic Filtering and Iterative Feature Refinement with Multi-Attention for Joint Video Super-Resolution and Deblurring (CVPR 2024) · better on 8 of 8 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
- Self-Supervised Deep Blind Video Super-Resolution (TPAMI 2024 · similarity 0.88)
- Deep Constrained Least Squares for Blind Image Super-Resolution (CVPR 2022 · similarity 0.88)
- Blind Super-Resolution With Iterative Kernel Correction (CVPR 2019 · similarity 0.87)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.87)
- A Dynamic Kernel Prior Model for Unsupervised Blind Image Super-Resolution (CVPR 2024 · similarity 0.86)
- Event-Enhanced Blurry Video Super-Resolution (AAAI 2025 · similarity 0.85)
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.
