Research map / Video / Video segmentation and tracking
Event-Guided Deblurring of Unknown Exposure Time Videos
Taewoo Kim, Jeongmin Lee, Lin Wang, Kuk-Jin Yoon
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 recurrent encoding (shared RNN cell) for event streams instead of 2D CNN (ablation) | event-guided deblurring network · Real-world event dataset | PSNR (dB): 34.45 -> 34.96 | Improved |
| add multi-scale loss (ablation) | network with recurrent encoding · Real-world event dataset | PSNR (dB): 34.96 -> 35.42 | Improved |
| add ETES module (ablation) | network with recurrent encoding · Real-world event dataset | PSNR (dB): 34.96 -> 35.77 | Improved |
| add proposed feature fusion module instead of concatenation (ablation) | network with RE, MS and ETES · Real-world event dataset | PSNR (dB): 36.02 -> 36.35 | Improved |
| add multi-scale loss to network with ETES and FF (ablation) | network with RE, ETES, FF · Real-world event dataset | PSNR (dB): 36.09 -> 36.35 | Improved |
| train with exposure ratios spanning m=9..15 (unknown exposure) rather than known exposure assumption (ablation) | Ours · GoPro unseen intervals | PSNR (dB): 26.52 -> 35.65 | Improved |
| add random noise to readout interval during training (ablation) | Ours · GoPro unseen intervals | PSNR (dB): 35.65 -> 36.77 | Improved |
| retrain LEDVDI on the real-world event data instead of using pretrained model | LEDVDI · Real-world event dataset | PSNR (dB): 23.73 -> 33.86 | Improved |
| use full Ours instead of Ours-light | Ours · Real-world event dataset | PSNR (dB): 34.92 -> 36.35 | Improved |
| add events to Nah et al. as input (retrained) | Nah et al. · Real-world event dataset | PSNR (dB): 28.81 -> 33.45 | Improved |
| use Ours instead of frame-based competitors on GoPro-15-1 | Ours · GoPro-15-1 | PSNR (dB): 29.66 -> 35.28 | Improved |
Closest papers by meaning
- Frequency-aware Event-based Video Deblurring for Real-World Motion Blur (CVPR 2024 · similarity 0.93)
- Motion Aware Event Representation-driven Image Deblurring (ECCV 2024 · similarity 0.93)
- Separation for Better Integration: Disentangling Edge and Motion in Event-based Deblurring (ICCV 2025 · similarity 0.93)
- Diffusion-based Event Generation for High-Quality Image Deblurring (CVPR 2025 · similarity 0.90)
- Latency Correction for Event-guided Deblurring and Frame Interpolation (CVPR 2024 · similarity 0.89)
- Motion Deblurring via Spatial-Temporal Collaboration of Frames and Events (AAAI 2024 · similarity 0.89)
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.
