Research map / Video / Video segmentation and tracking
Generalizing Event-Based Motion Deblurring in Real-World Scenarios
Xiang Zhang, Lei Yu, Wen Yang, Jianzhuang Liu, Gui-Song Xia
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 structure consistency loss L_SC to brightness consistency L_BC (ablation) | SAN self-supervised training · Ev-REDS (R=4) | PSNR: 19.95 -> 23.39 | Improved |
| train with L_SC alone without L_BC (ablation) | SAN self-supervised training · Ev-REDS (R=4) | PSNR: 18.95 (vs 23.39 with both) | Hurt |
| add temporal generalization self-distillation L_TG (ablation) | SAN self-supervised training · Ev-REDS (R=4) | PSNR: 23.39 -> 24.00 | Improved |
| add temporal generalization L_TG (ablation) | SAN self-supervised training · Ev-REDS (R=1) | PSNR: 21.77 -> 21.73 | No clear effect |
| add spatial generalization self-distillation L_SG (ablation) | SAN self-supervised training · Ev-REDS (R=1) | PSNR: 21.77 -> 23.46 | Improved |
| use both L_TG and L_SG together (ablation) | SAN self-supervised training · Ev-REDS (R=1 / R=4) | PSNR: R=1: 21.77 -> 24.12; R=4: 23.39 -> 23.95 | Improved |
| use native scale-aware network vs cascaded super-resolution plus deblurring | event deblurring with HR frames and LR events · Ev-REDS (R=4, color) | PSNR: 23.35 -> 23.95 | Improved |
| evaluate on large blur unseen in training | SAN vs EVDI · HS-ERGB large blur (color) | PSNR: 24.08 -> 25.41 | Improved |
Closest papers by meaning
- Motion Deblurring With Real Events (ICCV 2021 · similarity 0.94)
- Frequency-aware Event-based Video Deblurring for Real-World Motion Blur (CVPR 2024 · similarity 0.93)
- Learning Scale-Aware Spatio-temporal Implicit Representation for Event-based Motion Deblurring (ICML 2024 · similarity 0.92)
- Event-Based Motion Deblurring Using Task-Oriented 3D Gaussian Event Representations (CVPR 2026 · similarity 0.92)
- Motion-adaptive Transformer for Event-based Image Deblurring (AAAI 2025 · similarity 0.90)
- Motion Deblurring via Spatial-Temporal Collaboration of Frames and Events (AAAI 2024 · similarity 0.90)
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.
