Research map / 3D vision / Gaussian splatting
GP-4DGS: Probabilistic 4D Gaussian Splatting from Monocular Video via Variational Gaussian Processes
Mijeong Kim, Jungtaek Kim, Bohyung Han
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 GP-based motion prior and GP guidance loss | SoM · DyCheck all scenes | mPSNR: 17.09 -> 17.38 | Improved |
| add GP-based motion prior and GP guidance loss | SoM · DyCheck all scenes | mSSIM: 0.65 -> 0.65 | No clear effect |
| add GP-based motion prior | SoM · DyCheck challenging subset | mPSNR: 14.56 -> 15.02 | Improved |
| add GP-based motion prior | SoM · DyCheck challenging subset | mSSIM: 0.46 -> 0.46 | No clear effect |
| add GP-based motion prior | SoM · DyCheck 5 scenes | mLPIPS: 0.43 -> 0.41 | Improved |
| use GP extrapolation instead of linear extrapolation | linear extrapolation · periodic motion, last 5 frames | PSNR: 11.55 -> 17.62 | Improved |
| use GP extrapolation instead of linear extrapolation | linear extrapolation · non-periodic motion, last 5 frames | PSNR: 15.02 -> 15.27 | Improved |
| use GP-4DGS uncertainty instead of UA-4DGS | UA-4DGS · Uncertainty quantification (top 20 frames) | AUSE-MSE (x10^-2): 7.60 -> 7.22 | Improved |
| use GP-4DGS uncertainty instead of UA-4DGS | UA-4DGS · Uncertainty quantification (all frames) | AUSE-MSE (x10^-2): 8.62 -> 8.49 | Improved |
| use time-series (Chronos) inducing point initialization instead of random (ablation) | GP-4DGS · ELBO avg over 7 scenes | ELBO: 1.10 -> 1.53 | Improved |
| use time-series inducing point initialization instead of velocity KNN (ablation) | GP-4DGS · ELBO avg over 7 scenes | ELBO: 1.37 -> 1.53; apple 1.21 -> 1.19 | Mixed |
Closest papers by meaning
- 4D Gaussian Splatting in the Wild with Uncertainty-Aware Regularization (NeurIPS 2024 · similarity 0.92)
- Orientation-anchored Hyper-Gaussian for 4D Reconstruction from Casual Videos (NeurIPS 2025 · similarity 0.91)
- 1000+ FPS 4D Gaussian Splatting for Dynamic Scene Rendering (NeurIPS 2025 · similarity 0.90)
- Instant4D: 4D Gaussian Splatting in Minutes (NeurIPS 2025 · similarity 0.90)
- Fully Explicit Dynamic Gaussian Splatting (NeurIPS 2024 · similarity 0.90)
- DGS-LRM: Real-Time Deformable 3D Gaussian Reconstruction From Monocular Videos (NeurIPS 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.
