Research map / 3D vision / Gaussian splatting
HeroGS: Hierarchical Guidance for Robust 3D Gaussian Splatting under Sparse Views
Jiashu Li, Xumeng Han, Zhaoyang Wei, Zipeng Wang, Kuiran Wang, Guorong Li, Zhenjun Han, Jianbin Jiao
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 frame-interpolation pseudo-labels (image-level guidance) to FSGS (ablation) | FSGS · LLFF (2 views) | PSNR: 15.65 -> 16.91 | Improved |
| add FADP on top of pseudo-labels (ablation) | FSGS + VFI · LLFF (2 views) | PSNR: 16.91 -> 17.28 | Improved |
| add auxiliary GS field with frozen scale (co-pruning) (ablation) | FSGS + VFI + FADP · LLFF (2 views) | PSNR: 17.28 -> 17.93 | Improved |
| use two auxiliary GS fields (scale and rotation frozen) instead of one (ablation) | HeroGS · LLFF (2 views) | PSNR: 17.93 -> 18.78 | Improved |
| add post-freeze behavior to co-pruning (ablation) | HeroGS · LLFF (3 views) | PSNR: 21.16 -> 21.30 | Improved |
| use VFI instead of PerVFI as interpolator (ablation) | HeroGS without CPG · LLFF (3 views) | PSNR: 20.78 -> 20.99 | Improved |
| replace interpolated pseudo-labels with ground-truth images (ablation) | HeroGS without CPG · LLFF (3 views) | PSNR: 20.99 -> 21.09 | Improved |
| add CPG on top of interpolated pseudo-labels (ablation) | HeroGS · LLFF (3 views) | PSNR: 20.99 -> 21.30 | Improved |
| increase interpolation factor from S=2 to S=4 or more (ablation) | HeroGS · LLFF | PSNR/SSIM/LPIPS | Improved |
| combine two levels instead of one (average) (ablation) | HeroGS · LLFF (2 views) | PSNR: 16.78 -> 17.68 | Improved |
| combine all three levels instead of two (average) (ablation) | HeroGS · LLFF (2 views) | PSNR: 17.68 -> 18.78 | Improved |
| HeroGS vs DropGaussian at 6 views | HeroGS · LLFF (6 views) | LPIPS: 0.117 (DropGaussian) vs 0.135 (HeroGS) | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DropGaussian — DropGaussian: Structural Regularization for Sparse-view Gaussian Splatting (CVPR 2025) · better on 4 of 6 shared results
- CoR-GS — CoR-GS: Sparse-View 3D Gaussian Splatting via Co-Regularization (ECCV 2024) · better on 3 of 3 shared results
- DRGS — DR-GS: Physically-Based Deformable and Relightable 2D Gaussians (ECCV 2026) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Efficient Scene Modeling via Structure-Aware and Region-Prioritized 3D Gaussians (TPAMI 2025 · similarity 0.93)
- GaussianSpa: An "Optimizing-Sparsifying" Simplification Framework for Compact and High-Quality 3D Gaussian Splatting (CVPR 2025 · similarity 0.92)
- ReorgGS: Equivalent Distribution Reorganization for 3D Gaussian Splatting (NeurIPS 2026 · similarity 0.92)
- LP-3DGS: Learning to Prune 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.92)
- Steepest Descent Density Control for Compact 3D Gaussian Splatting (CVPR 2025 · similarity 0.92)
- IBGS: Image-Based Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
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.
