Research map / Video / Video segmentation and tracking
Alignment-free HDR Deghosting with Semantics Consistent Transformer
Steven Tel, Zongwei Wu, Yulun Zhang, Barthélémy Heyrman, Cédric Demonceaux, Radu Timofte, Dominique Ginhac
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 G-SAB spatial attention block to baseline (ablation) | SCTNet · Kalantari test | mu-PSNR: 43.42 -> 44.04 | Improved |
| add S-CAB cross-frame channel attention on top of G-SAB (ablation) | SCTNet · Kalantari test | mu-PSNR: 44.04 -> 44.12 | Improved |
| add S-CAB on top of G-SAB (ablation) | SCTNet · Kalantari test | PSNR (linear): 41.89 -> 41.71 | Hurt |
| add skip connection from reference shallow feature (ablation) | SCTNet · Kalantari test | mu-PSNR / PSNR: 44.12 -> 44.49 (mu-PSNR); 41.71 -> 42.29 (PSNR) | Improved |
| SCTNet versus HDR-Transformer | SCTNet · Kalantari test | mu-PSNR: 44.32 -> 44.49 | Improved |
| SCTNet versus HDR-Transformer on proposed dataset | SCTNet · Proposed HDR dataset | l-PSNR: 46.35 -> 47.51 | Improved |
| alignment-free attention design versus HDR-Transformer | SCTNet · 1000x1500 input | GMACs: 981.81 -> 293.77 | Improved |
| SCTNet versus convolutional methods on runtime | SCTNet · Quadro P6000, 1000x1500 | Speed (s): 0.29 (HDR-GAN) vs 7.14 | Hurt |
Closest papers by meaning
- A Unified HDR Imaging Method With Pixel and Patch Level (CVPR 2023 · similarity 0.92)
- Towards Real-World HDR Video Reconstruction: A Large-Scale Benchmark Dataset and A Two-Stage Alignment Network (CVPR 2024 · similarity 0.89)
- HDRFlow: Real-Time HDR Video Reconstruction with Large Motions (CVPR 2024 · similarity 0.89)
- Self-Supervised High Dynamic Range Imaging with Multi-Exposure Images in Dynamic Scenes (ICLR 2024 · similarity 0.87)
- Single Shot HDR Recovery via a Video Diffusion Prior (NeurIPS 2026 · similarity 0.87)
- High Dynamic Range Novel View Synthesis with Single Exposure (ICML 2025 · similarity 0.87)
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.
