Research map / Image generation / Super-resolution
VoDaSuRe: A Large-Scale Dataset Revealing Domain Shift in Volumetric Super-Resolution
August Leander Høeg, Sophia Wiinberg Bardenfleth, Hans Martin Kjer, Tim Bjørn Dyrby, Vedrana Andersen Dahl, Anders Bjorholm Dahl
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 |
|---|---|---|---|
| train on real registered LR instead of downsampled LR (RRDBNet3D, 4x) | SR models · VoDaSuRe | PSNR: 19.08 -> 16.22 | Hurt |
| use real scanned LR instead of downsampled LR at 2x (RRDBNet3D) | SR models · VoDaSuRe | PSNR: 25.50 -> 18.25 | Hurt |
| use downsampled LR on clinical CT (RRDBNet3D, 2x) vs VoDaSuRe downsampled | SR models · CTSpine1K vs VoDaSuRe (downsampled) | PSNR: 39.88 -> 25.50 | Hurt |
| train on downsampled LR, test on registered real LR (RRDBNet3D, 4x) | RRDBNet3D · VoDaSuRe (registered) 4x | PSNR: 16.22 -> 14.94 | Hurt |
| train on downsampled LR, test on registered real LR (RRDBNet3D, 2x) | RRDBNet3D · VoDaSuRe (registered) 2x | PSNR: 18.25 -> 16.74 | Hurt |
| add deliberate misregistration between downsampled LR and HR (ablation) | RRDBNet3D · VoDaSuRe Elm 4x | sharpness (visual) | No clear effect |
| add LPIPS loss (lambda 0.02) to L1 (ablation) | RRDBNet3D · VoDaSuRe 4x | visual texture | Mixed |
| downsample both HR and real LR by 2x to get 4x difference (ablation) | RRDBNet3D and MTVNet · VoDaSuRe Elm / Wood | PSNR | No clear effect |
| exclude test materials (elm, cypress) from training (ablation) | RRDBNet3D and MTVNet · VoDaSuRe 2x | PSNR | No clear effect |
| use real LR scans: total variation of predictions vs downsampled-LR predictions | RRDBNet3D · VoDaSuRe | total variation | Hurt |
Closest papers by meaning
- MedGSSR: Generalizable Medical Image Super-Resolution 3D Reconstruction via Hierarchical Feed-forward Gaussian Splatting (ECCV 2026 · similarity 0.86)
- Pairwise Distance Distillation for Unsupervised Real-World Image Super-Resolution (ECCV 2024 · similarity 0.86)
- CycleINR: Cycle Implicit Neural Representation for Arbitrary-Scale Volumetric Super-Resolution of Medical Data (CVPR 2024 · similarity 0.85)
- Learning General-purpose Biomedical Volume Representations using Randomized Synthesis (ICLR 2025 · similarity 0.85)
- Navigating Beyond Dropout: An Intriguing Solution towards Generalizable Image Super Resolution (CVPR 2024 · similarity 0.84)
- Towards Realistic Data Generation for Real-World Super-Resolution (ICLR 2025 · similarity 0.84)
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.
