Research map / 3D vision / Face and shape
NAB: Neural Adaptive Binning for Sparse-View CT reconstruction
Wangduo Xie, Matthew B. Blaschko
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 |
|---|---|---|---|
| replace random Fourier encoding with NAB | INR_f · CaCO3 (16-view) | PSNR (dB): 29.01 -> 43.61 | Improved |
| replace random Fourier encoding with NAB | INR_f · CaCO3 (12-view) | PSNR (dB): 25.08 -> 34.72 | Improved |
| replace random Fourier encoding with NAB | INR_f · Workpieces (12-view) | PSNR (dB): 27.37 -> 32.76 | Improved |
| replace random Fourier encoding with NAB | INR_f · Workpieces (14-view) | PSNR (dB): 34.03 -> 35.23 | Improved |
| use NAB with 4-layer FCN instead of 7-layer wider INR_l2 | INR_l2 · CaCO3 (16-view) | PSNR (dB): 38.89 -> 43.61 | Improved |
| train 29,990 epochs instead of 19,990 | NAB · CaCO3 (14-view) | PSNR (dB): 36.09 -> 40.07 | Improved |
| add trainable bin center location (ablation) | NAB · CaCO3 (16-view) | PSNR (dB): 29.82 -> 43.61 | Improved |
| add trainable bin side length (ablation) | NAB · CaCO3 (16-view) | PSNR (dB): 30.82 -> 43.61 | Improved |
| add bin rotation (ablation) | NAB · CaCO3 (16-view) | PSNR (dB): 37.50 -> 43.61 | Improved |
| add trainable steepness (ablation) | NAB · CaCO3 (16-view) | PSNR (dB): 42.91 -> 43.61 | Improved |
| add trainable bin height (ablation) | NAB · CaCO3 (16-view) | PSNR (dB): 43.37 -> 43.61 | No clear effect |
| use NAB instead of Instant-NGP | Instant-NGP · CaCO3 (12-view) | SSIM: 0.911 -> 0.888 | Hurt |
Closest papers by meaning
- Near-Exact Recovery for Tomographic Inverse Problems via Deep Learning (ICML 2022 · similarity 0.86)
- QN-Mixer: A Quasi-Newton MLP-Mixer Model for Sparse-View CT Reconstruction (CVPR 2024 · similarity 0.85)
- Spatiotemporal-Aware Neural Fields for Dynamic CT Reconstruction (AAAI 2025 · similarity 0.84)
- Structure-Aware Sparse-View X-ray 3D Reconstruction (CVPR 2024 · similarity 0.84)
- C^2RV: Cross-Regional and Cross-View Learning for Sparse-View CBCT Reconstruction (CVPR 2024 · similarity 0.83)
- ND-SDF: Learning Normal Deflection Fields for High-Fidelity Indoor Reconstruction (ICLR 2025 · similarity 0.83)
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.
