Research map / 3D vision / Gaussian splatting
DualSplat: Robust 3D Gaussian Splatting via Pseudo-Mask Bootstrapping from Reconstruction Failures
Xu Wang, Zhiru Wang, Shiyun Xie, Chengwei Pan, Yisong Chen
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 pseudo-masks to 3DGS (ablation) | 3DGS · NeRF On-the-go (mean) | PSNR: 19.043 -> 22.604 (+3.56 dB) | Improved |
| add pseudo-masks to 3DGS with delayed densification (ablation) | DD · NeRF On-the-go (mean) | PSNR: 20.820 -> 22.899 | Improved |
| add delayed densification to 3DGS (ablation) | 3DGS · NeRF On-the-go (mean) | PSNR: 19.043 -> 20.820 | Improved |
| add MLP mask predictor with pseudo-mask supervision (ablation) | DD+PM · NeRF On-the-go (mean) | PSNR: 22.899 -> 23.262 (+0.37 dB) | Improved |
| add robustness loss to the MLP (ablation) | DD+MLP · NeRF On-the-go (mean) | PSNR: 22.902 -> 23.262 | Improved |
| add pseudo-mask supervision to the MLP (ablation) | DD+MLP · NeRF On-the-go (mean) | PSNR: 23.122 -> 23.262 | Improved |
| add depth regularization (ablation) | DD+MLP · NeRF On-the-go (mean) | PSNR: 23.262 -> 23.421 | Improved |
| add depth regularization (ablation) | DD+MLP · NeRF On-the-go (mean) | LPIPS: 0.088 -> 0.088 | No clear effect |
| use FiT3D features instead of DINOv2 for the mask filter (ablation) | mask filter · Transient mask quality | IoU: 0.744 -> 0.835 | Improved |
| use FiT3D features instead of ResNet50 (ablation) | mask filter · Transient mask quality | IoU: 0.616 -> 0.835 | Improved |
| MLP-refined masks instead of FiT3D filter alone (ablation) | mask filter · Transient mask quality | IoU: 0.835 -> 0.858 | Improved |
| DualSplat vs RobustSplat on transient regions | DualSplat · RobustNeRF Crab1 (inside transient regions) | PSNR: 32.78 -> 33.04 | Improved |
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.
- DeSplat — DeSplat: Decomposed Gaussian Splatting for Distractor-Free Rendering (CVPR 2025) · better on 4 of 4 shared results
- WildGaussians — WildGaussians: 3D Gaussian Splatting In the Wild (NeurIPS 2024) · better on 4 of 4 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
- Robust Neural Rendering in the Wild with Asymmetric Dual 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.92)
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.91)
- HybridGS: Decoupling Transients and Statics with 2D and 3D Gaussian Splatting (CVPR 2025 · similarity 0.90)
- Learning Efficient Fuse-and-Refine for Feed-Forward 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.90)
- MaskGaussian: Adaptive 3D Gaussian Representation from Probabilistic Masks (CVPR 2025 · similarity 0.90)
- SplatFormer: Point Transformer for Robust 3D Gaussian Splatting (ICLR 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.
