Research map / Video / Video segmentation and tracking
Frames2Residual: Spatiotemporal Decoupling for Self-Supervised Video Denoising
Mingjie Ji, Zhan Shi, Kailai Zhou, Zixuan Fu, Xun Cao
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 Stage 1 blind temporal estimator (ablation) | image denoiser baseline · DAVIS (sigma=30) | PSNR: 33.47 -> 34.36 | Improved |
| add Stage 2 spatial refiner on top of Stage 1 (ablation) | F2R Stage 1 · DAVIS (sigma=30) | PSNR: 34.36 -> 35.56 | Improved |
| add Stage 2 without Stage 1 anchor (ablation) | image denoiser baseline · DAVIS (sigma=30) | PSNR: 33.47 -> 33.47 | No clear effect |
| add Stage 2 on top of Stage 1 (ablation) | F2R Stage 1 · CRVD (ISO 25600) | PSNR: 41.88 -> 42.82 | Improved |
| add Stage 2 on top of Stage 1 (ablation) | F2R Stage 1 · Set8 (sigma=30) | PSNR: 32.70 -> 33.82 | Improved |
| add flow-guided FAAM to U-Net (ablation) | U-Net baseline · DAVIS (sigma=30, Stage 1) | PSNR: 25.06 -> 28.22 | Improved |
| add joint inputs for residual prediction (ablation) | U-Net baseline · DAVIS (sigma=30, Stage 1) | PSNR: 25.06 -> 33.59 | Improved |
| add FAAM on top of joint inputs (ablation) | Stage 1 U-Net · DAVIS (sigma=30, Stage 1) | PSNR: 33.59 -> 34.36 | Improved |
| use FDAM in Stage 2 instead of FAAM (ablation) | F2R · DAVIS (sigma=30) | PSNR: 35.37 -> 35.56 | Improved |
| use FAAM in Stage 1 instead of FDAM (ablation) | F2R · DAVIS (sigma=30) | PSNR: 35.53 -> 35.56 | Improved |
| increase temporal window from T=3 to T=9 (ablation) | F2R Stage 1 · DAVIS (sigma=30) | PSNR: 33.96 -> 34.36 | Improved |
| increase temporal window from T=9 to T=11 (ablation) | F2R Stage 1 · DAVIS (sigma=30) | PSNR: 34.36 -> 34.39 | No clear effect |
Closest papers by meaning
- Learning Continuous Spatiotemporal Implicit Neural Fields for Unsupervised Video Denoising (TPAMI 2026 · similarity 0.91)
- Exploring Efficient Asymmetric Blind-Spots for Self-Supervised Denoising in Real-World Scenarios (CVPR 2024 · similarity 0.89)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 2025 · similarity 0.88)
- Spatiotemporal Blind-Spot Network with Calibrated Flow Alignment for Self-Supervised Video Denoising (AAAI 2025 · similarity 0.87)
- TD²-Net: Toward Denoising and Debiasing for Video Scene Graph Generation (AAAI 2024 · similarity 0.87)
- One-Step Diffusion for Detail-Rich and Temporally Consistent Video Super-Resolution (NeurIPS 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.
