Research map / Video / Video segmentation and tracking
Learning to Extract a Video Sequence From a Single Motion-Blurred Image
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 |
|---|---|---|---|
| Use the proposed middle-frame predictor instead of Nah's network | Middle frame prediction · Nah test set (45%) | PSNR (dB): 30.52 -> 32.20 | Helped |
| Use the proposed middle-frame predictor instead of Nah's network | Middle frame prediction · Our testset | PSNR (dB): 28.19 -> 29.02 | Helped |
| Use the proposed middle-frame predictor instead of Nah's network | Middle frame prediction · Nah original test images | PSNR (dB): 28.48 -> 26.98 | Hurt |
| Use the proposed middle-frame network instead of Nah's | Middle frame prediction · 320P | execution time: 2.43 -> 0.24 | Helped |
| Train each frame independently with a direct target-frame loss (ablation) | Seven-frame reconstruction · Single blurred image to seven frames | visual reconstruction quality | Hurt |
| Replace pairwise ordering-invariant losses with the global ordering-invariant loss (ablation) | Seven-frame reconstruction · Single blurred image to seven frames | visual reconstruction quality | Hurt |
| Use pairwise ordering-invariant losses without sequential ordering (ablation) | Seven-frame reconstruction · Single blurred image to seven frames | visual reconstruction quality | Mixed |
| Feed the predicted middle frame to pairwise networks (ablation) | Pairwise Frame Reconstruction · Single blurred image to seven frames | visual reconstruction quality | Helped |
| Sequentially order pairwise outputs using previous network outputs (ablation) | Seven-frame reconstruction · Single blurred image to seven frames | visual reconstruction quality | Helped |
| Use the ground-truth middle frame during non-middle-frame training (ablation) | Sequential pairwise training · Single blurred image to seven frames | visual reconstruction quality | Helped |
| Use the ground-truth middle frame during non-middle-frame training (ablation) | Sequential pairwise training · Training convergence | convergence speed | Helped |
| Share parameters for pairwise networks predicting frames 1, 2, 6, and 7 (ablation) | Seven-frame reconstruction · Single blurred image to seven frames | visual reconstruction quality | No clear effect |
Closest papers by meaning
- Bringing Alive Blurred Moments (CVPR 2019 · similarity 0.92)
- Learning to Synthesize Motion Blur (CVPR 2019 · similarity 0.89)
- Deep Video Deblurring for Hand-Held Cameras (CVPR 2017 · similarity 0.89)
- Motion-blurred Video Interpolation and Extrapolation (AAAI 2021 · similarity 0.88)
- Efficient Multi-scale Network with Learnable Discrete Wavelet Transform for Blind Motion Deblurring (CVPR 2024 · similarity 0.86)
- Parameterized Blur Kernel Prior Learning for Local Motion Deblurring (CVPR 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.