Research map / Video / Video segmentation and tracking
EventHDR: From Event to High-Speed HDR Videos and Beyond
Yunhao Zou, Ying Fu, Tsuyoshi Takatani, Yinqiang 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 key frame guidance to uni-directional RNN (ablation) | Ours · EventHDR real test | RMSE/TC: RMSE 0.1105 -> 0.1039; TC 0.3125 -> 0.2524 | Improved |
| use uni-directional recurrence instead of sliding window (ablation) | Ours · EventHDR real test | RMSE: 0.1153 -> 0.1105 | Improved |
| key frame interval K=5 instead of K=10 (ablation) | Ours · EventHDR real test | RMSE/SSIM: RMSE 0.1132 -> 0.1039; SSIM 0.5647 -> 0.5975 | Improved |
| key frame interval K=5 instead of K=1 (ablation) | Ours · EventHDR real test | RMSE: 0.1118 -> 0.1039 | Improved |
| add PCD deformable alignment instead of none (ablation) | Ours · EventHDR real test | RMSE/SSIM: RMSE 0.1382 -> 0.1039; SSIM 0.4657 -> 0.5975 | Improved |
| PCD alignment instead of optical flow (SpyNet) (ablation) | Ours · EventHDR real test | RMSE: 0.1279 -> 0.1039 | Improved |
| local attention fusion instead of addition (ablation) | Ours · EventHDR real test | RMSE: 0.2159 -> 0.1039 | Improved |
| local attention instead of global temporal-spatial attention (ablation) | Ours · EventHDR real test | RMSE/Params: RMSE 0.1063 -> 0.1039; Params 308.03K -> 40.18K | Mixed |
| add temporal consistency loss (ablation) | Ours · EventHDR real test | RMSE/TC: RMSE 0.1118 -> 0.1039; TC 0.3072 -> 0.2524 | Improved |
| add LPIPS loss (ablation) | Ours · EventHDR real test | RMSE/SSIM: RMSE 0.4536 -> 0.1039; SSIM 0.2045 -> 0.5975 | Improved |
| add l1 loss (ablation) | Ours · EventHDR real test | RMSE/SSIM: RMSE 0.1294 -> 0.1039; SSIM 0.4260 -> 0.5975 | Improved |
| train on real paired event/HDR data (Case 4) instead of simulated events from LDR (Case 1) (ablation) | Ours · EventHDR real test | RMSE/SSIM: RMSE 0.2032 -> 0.1039; SSIM 0.3191 -> 0.5975 | Improved |
Closest papers by meaning
- Learning To Reconstruct High Speed and High Dynamic Range Videos From Events (CVPR 2021 · similarity 0.95)
- Learning Event Guided High Dynamic Range Video Reconstruction (CVPR 2023 · similarity 0.92)
- EvHDR-GS: Event-guided HDR Video Reconstruction with 3D Gaussian Splatting (AAAI 2025 · similarity 0.90)
- Towards Real-World HDR Video Reconstruction: A Large-Scale Benchmark Dataset and A Two-Stage Alignment Network (CVPR 2024 · similarity 0.88)
- HDRFlow: Real-Time HDR Video Reconstruction with Large Motions (CVPR 2024 · similarity 0.88)
- E2HQV: High-Quality Video Generation from Event Camera via Theory-Inspired Model-Aided Deep Learning (AAAI 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.
