Research map / Video / Video segmentation and tracking
Recurrent Neural Networks With Intra-Frame Iterations for Video Deblurring
Check 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 |
|---|---|---|---|
| Add intra-frame hidden-state iterations (ablation) | Single-cell IFI-RNN · GOPRO test (downsampled) | PSNR: 28.79 -> 29.07 | Helped |
| Increase single-cell hidden-state iterations from three to four (ablation) | IFI-RNN · GOPRO test (downsampled) | PSNR: 29.07 -> 29.06 | Hurt |
| Add a second RNN cell for separate hidden-state initialization and updating (ablation) | IFI-RNN · GOPRO test (downsampled) | PSNR: 28.79 -> 29.72 | Helped |
| Add stochastic regularization favoring shorter computation paths (ablation) | Three-iteration dual-cell IFI-RNN · GOPRO test (downsampled) | PSNR: 29.80 -> 29.97 | Helped |
| Add stochastic regularization favoring shorter computation paths (ablation) | Four-iteration single-cell IFI-RNN · GOPRO test (downsampled) | PSNR: 29.06 -> 29.16 | Helped |
| Add stochastic regularization favoring shorter computation paths (ablation) | Four-iteration dual-cell IFI-RNN · GOPRO test (downsampled) | PSNR: 29.82 -> 29.93 | Helped |
| Add stochastic regularization favoring shorter computation paths (ablation) | Single-cell IFI-RNN · Dataset from [34] | PSNR: 30.07 -> 30.10 | Helped |
| Add stochastic regularization to the three-iteration dual-cell model rather than using four-iteration regularized inference (ablation) | Dual-cell IFI-RNN · Dataset from [34] | PSNR: 30.73 -> 30.80 | Helped |
| Use more hidden-state iterations (ablation) | IFI-RNN · GOPRO test (downsampled) | PSNR: C1H3 29.07 -> C1H4 29.06; C2H3 29.80 -> C2H4 29.82 | Mixed |
| Use stochastic regularization rather than fixed-iteration training (ablation) | IFI-RNN · Dataset from [34] | PSNR: C2H3-reg 30.80; C2H4-reg 30.73 | Mixed |
Closest papers by meaning
- Deep Recurrent Neural Network with Multi-Scale Bi-directional Propagation for Video Deblurring (AAAI 2022 · similarity 0.90)
- Scale-Recurrent Network for Deep Image Deblurring (CVPR 2018 · similarity 0.90)
- Cascaded Deep Video Deblurring Using Temporal Sharpness Prior (CVPR 2020 · similarity 0.88)
- Dynamic Scene Deblurring Using Spatially Variant Recurrent Neural Networks (CVPR 2018 · similarity 0.88)
- Multi-Scale Memory-Based Video Deblurring (CVPR 2022 · similarity 0.87)
- Event-Based Frame Interpolation With Ad-Hoc Deblurring (CVPR 2023 · similarity 0.87)
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.