Research map / Video / Video segmentation and tracking
Efficient Spatio-Temporal Recurrent Neural Network for Video Deblurring
Zhihang Zhong, Ye Gao, Yinqiang Zheng, Bo Zheng
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 neighboring-frame high-level feature fusion (simple concatenation) (ablation) | baseline RNN B9C110 · GOPRO | PSNR (dB): 30.29 -> 30.46 | Improved |
| add RDB cell instead of residual block (ablation) | baseline RNN B9C110 · GOPRO | PSNR (dB): 30.29 -> 30.51 | Improved |
| add GSA module on top of fusion and RDB cell (ablation) | ESTRNN · GOPRO | PSNR (dB): 30.55 -> 30.79 | Improved |
| add fusion strategy to RDB cell plus GSA (ablation) | ESTRNN · GOPRO | PSNR (dB): 30.69 -> 30.79 | Improved |
| add fusion to the RDB cell model without GSA (ablation) | ESTRNN · GOPRO | PSNR (dB): 30.51 -> 30.55 | Improved |
| increase RDB blocks from 3 to 15 (C80) (ablation) | ESTRNN · GOPRO | PSNR (dB): 29.74 -> 31.27 | Improved |
| increase RDB blocks from 9 to 12 (C80) (ablation) | ESTRNN · GOPRO | PSNR (dB): 30.79 -> 31.03 | Improved |
| use B15C80 rather than wider B9C90 at similar cost (ablation) | ESTRNN · GOPRO | PSNR (dB): 31.07 -> 31.27 | Improved |
| add 2 future frames to 2 past frames to reach 2 past and 2 future (ablation) | ESTRNN B9C80 · GOPRO | PSNR (dB): 30.57 -> 30.79 | Improved |
| use 3 past frames only instead of 2 past plus 2 future (ablation) | ESTRNN B9C80 · GOPRO | PSNR (dB): 30.69 vs 30.79 | No clear effect |
| increase neighbors from F2P2 to F3P3 (ablation) | ESTRNN B9C80 · GOPRO | PSNR (dB): 30.79 -> 30.82 | No clear effect |
| use 1 past frame instead of 1 past and 1 future frame (F1P1) (ablation) | ESTRNN B9C80 · GOPRO | PSNR (dB): 30.54 -> 30.58 | No clear effect |
Closest papers by meaning
- Real-World Video Deblurring: A Benchmark Dataset and an Efficient Recurrent Neural Network (IJCV 2022 · similarity 0.96)
- Online Video Deblurring via Dynamic Temporal Blending Network (ICCV 2017 · similarity 0.91)
- Learning Efficient Deep Discriminative Spatial and Temporal Networks for Video Deblurring (TPAMI 2025 · similarity 0.90)
- Gated Spatio-Temporal Attention-Guided Video Deblurring (CVPR 2021 · similarity 0.90)
- Blur-aware Spatio-temporal Sparse Transformer for Video Deblurring (CVPR 2024 · similarity 0.89)
- Learning Scale-Aware Spatio-temporal Implicit Representation for Event-based Motion Deblurring (ICML 2024 · 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.
