Research map / Video / Video segmentation and tracking
CMTA: Cross-Modal Temporal Alignment for Event-guided Video Deblurring
Taewoo Kim, Hoonhee Cho, 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 the CRIFE module (ablation) | CMTA baseline (no ECITFA) · GoPro | PSNR: 34.65 -> 35.00 | Improved |
| add the ECITFA module (ablation) | CMTA baseline (no CRIFE) · GoPro | PSNR: 34.65 -> 36.34 | Improved |
| add CRIFE to the model with ECITFA (ablation) | CMTA · GoPro | PSNR: 36.34 -> 36.55 | Improved |
| add pixel-wise dynamic filters in alignment (ablation) | ECITFA · GoPro | PSNR: 36.18 -> 36.55 | Improved |
| use cross-attention instead of ResBlocks for feature aggregation in alignment (ablation) | ECITFA · GoPro | PSNR: 36.29 -> 36.55 | Improved |
| use cascaded temporal feature alignment (ablation) | ECITFA · GoPro | PSNR: 36.08 -> 36.55 | Improved |
| concatenate RGB and event features for the query in CRIFE (ablation) | CRIFE · GoPro | PSNR: 36.04 -> 36.55 | Improved |
| use recurrent structure in CRIFE (ablation) | CRIFE · GoPro | PSNR: 35.99 -> 36.55 | Improved |
| increase input frames from 5 to 7 | CMTA · GoPro | PSNR: 36.55 -> 36.78 | Improved |
| add event guidance over best frame-based video deblurring | Shift-Net vs CMTA-5 · GoPro | PSNR: 35.49 -> 36.55 | Improved |
Later papers that compared with this method
3 later accepted papers report CMTA-7 as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Event-driven Motion Deblurring via Trajectory-based Kernel Reconstruction (ECCV 2026 · better on 1 of 1 shared result)
- Separation for Better Integration: Disentangling Edge and Motion in Event-based Deblurring (ICCV 2025 · better on 1 of 4 shared results)
- EVDM: Event-based Real-world Video Deblurring with Mamba (ICCV 2025 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Frequency-aware Event-based Video Deblurring for Real-World Motion Blur (CVPR 2024 · similarity 0.95)
- Event-Guided Deblurring of Unknown Exposure Time Videos (ECCV 2022 · similarity 0.92)
- Time-Specialized Event-Image Alignment for Blur-to-Video Decomposition (CVPR 2026 · similarity 0.91)
- Motion Deblurring via Spatial-Temporal Collaboration of Frames and Events (AAAI 2024 · similarity 0.90)
- Event-Enhanced Blurry Video Super-Resolution (AAAI 2025 · similarity 0.89)
- Latency Correction for Event-guided Deblurring and Frame Interpolation (CVPR 2024 · similarity 0.88)
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.
