Research map / 3D vision / Gaussian splatting
3DGS-HPC: Distractor-free 3D Gaussian Splatting with Hybrid Patch-wise Classification
Jiahao Chen, Yipeng Qin, Ganlong Zhao, Xin Li, Wenping Wang, Guanbin Li
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 patch-wise partition instead of no partition (pixel-wise) (ablation) | 3DGS-HPC (DINOv2) · On-the-go (Avg.) | PSNR: 22.95 -> 23.86 | Improved |
| use patch-wise partition instead of SAM2 region grouping (ablation) | 3DGS-HPC (DINOv2) · RobustNeRF (Avg.) | PSNR: 29.45 -> 29.81 | Improved |
| use patch-wise partition instead of superpixel grouping (ablation) | 3DGS-HPC (DINOv2) · On-the-go (Avg.) | PSNR: 23.29 -> 23.86 | Improved |
| use patch-wise partition instead of feature clustering (ablation) | 3DGS-HPC (DINOv2) · On-the-go (Avg.) | PSNR: 23.71 -> 23.86 | Improved |
| add photometric-guided percentile threshold to perceptual classification (hybrid metric) versus perceptual GMM alone (ablation) | 3DGS-HPC (DINOv2) · On-the-go (Avg.) | PSNR: 23.40 -> 23.86 | Improved |
| naively combine photometric GMM and perceptual GMM (ablation) | 3DGS-HPC (DINOv2) · On-the-go (Avg.) | PSNR: 23.86 -> 23.06 | Hurt |
| use perceptual metric alone instead of photometric metric alone (ablation) | 3DGS-HPC (DINOv2) · RobustNeRF (Avg.) | PSNR: 29.82 -> 29.13 | Hurt |
| vary patch size away from 16 (below 8 or above 16) (ablation) | 3DGS-HPC · RobustNeRF and On-the-go | PSNR / SSIM | Hurt |
| use too few or too many warm-up steps (ablation) | 3DGS-HPC · RobustNeRF and On-the-go | PSNR | Hurt |
| classify patches directly instead of refining pixel-wise results with patches (ablation) | 3DGS-HPC · RobustNeRF (Avg.) | training hours: 2.53 -> 0.22 hrs; PSNR 29.74 -> 29.81 | Improved |
| add HPC to native 3DGS | 3DGS · RobustNeRF dataset avg | PSNR: 26.46 -> 29.87 (ResNet) | Improved |
| misclassify 10% of patch-level GT masks (ablation) | 3DGS with GT masks · RobustNeRF | PSNR: 29.93 -> 29.35 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HybridGS — HybridGS: Decoupling Transients and Statics with 2D and 3D Gaussian Splatting (CVPR 2025) · better on 3 of 3 shared results
- WildGaussians — WildGaussians: 3D Gaussian Splatting In the Wild (NeurIPS 2024) · better on 2 of 2 shared results
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
- LP-3DGS: Learning to Prune 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.91)
- Efficient Decoupled Feature 3D Gaussian Splatting via Hierarchical Compression (CVPR 2025 · similarity 0.91)
- FHGS: Feature-Homogenized Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- 3DGS-Enhancer: Enhancing Unbounded 3D Gaussian Splatting with View-consistent 2D Diffusion Priors (NeurIPS 2024 · similarity 0.91)
- HybridGS: Decoupling Transients and Statics with 2D and 3D Gaussian Splatting (CVPR 2025 · similarity 0.90)
- GaussianSpa: An "Optimizing-Sparsifying" Simplification Framework for Compact and High-Quality 3D Gaussian Splatting (CVPR 2025 · similarity 0.90)
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.
