Research map / 3D vision / Gaussian splatting
Active View Selection with Perturbed Gaussian Ensemble for Tomographic Reconstruction
Yulun Wu, Ruyi Zha, Wei Cao, Yingying Li, Yuanhao Cai, Yaoyao Liu
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 |
|---|---|---|---|
| use SSIM variance instead of L1 error as uncertainty metric (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.644 -> 34.078 | Improved |
| use SSIM variance instead of PSNR as uncertainty metric (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.390 -> 34.078 | Improved |
| stochastic density scaling instead of dropout (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 32.718 -> 34.078 | Improved |
| stochastic density scaling instead of position jittering (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 34.061 -> 34.078 | No clear effect |
| perturb only 10% lowest-density Gaussians instead of 20% (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.589 -> 34.078 | Improved |
| perturb 10% instead of 5% lowest-density Gaussians (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.680 -> 34.078 | Improved |
| ensemble size 10 instead of 40 (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.670 -> 34.078 | Improved |
| ensemble size 10 instead of 5 (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.952 -> 34.078 | Improved |
| scaling amplitude 0.5 instead of 1.0 (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 33.338 -> 34.078 | Improved |
| scaling amplitude 0.1/0.3 vs 0.5 (ablation) | Perturbed Gaussian Ensemble · Synthetic CT 24-view | 3D PSNR (dB): 34.040 / 34.057 vs 34.078 | No clear effect |
| Perturbed Gaussian Ensemble vs FisherRF | Radiative GS · Synthetic CT 24-view | 3D PSNR (dB): 33.347 -> 34.078 | Improved |
| Perturbed Gaussian Ensemble vs FisherRF | Radiative GS · Real-world CT 24-view | 3D PSNR (dB): 36.205 -> 36.399 | Improved |
Closest papers by meaning
- R$^2$-Gaussian: Rectifying Radiative Gaussian Splatting for Tomographic Reconstruction (NeurIPS 2024 · similarity 0.91)
- Spherical Bayesian Experimental Design for Active View Selection in 3D Gaussian Splatting (NeurIPS 2026 · similarity 0.90)
- Metropolis-Hastings Sampling for 3D Gaussian Reconstruction (NeurIPS 2025 · similarity 0.88)
- DropGaussian: Structural Regularization for Sparse-view Gaussian Splatting (CVPR 2025 · similarity 0.87)
- S2Gaussian: Sparse-View Super-Resolution 3D Gaussian Splatting (CVPR 2025 · similarity 0.86)
- DDGS-CT: Direction-Disentangled Gaussian Splatting for Realistic Volume Rendering (NeurIPS 2024 · similarity 0.86)
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.
