Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian-Inspired Space-Time Superpixels
Kent Gauen, Stanley Chan
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 |
|---|---|---|---|
| Control superpixel count with the temporally coherent split term instead of only the merge alpha hyperparameter (ablation) | BIST · SegTrackv2 (1065 superpixels) | SZV: 161.4 -> 136.4 | Improved |
| Control superpixel count with the temporally coherent split term instead of alpha (ablation) | BIST · SegTrackv2 (1065 superpixels) | TEX (%): 28.9 -> 31.9 | Improved |
| Control superpixel count with the temporally coherent split term instead of alpha (ablation) | BIST · SegTrackv2 (1065 superpixels) | UE-3D (x1e-3): 9.31 -> 8.74 | Improved |
| Control superpixel count with the temporally coherent split term instead of alpha (ablation) | BIST · SegTrackv2 (1300 superpixels) | Pooling (dB): 27.91 -> 27.99 (alpha higher) | No clear effect |
| Control superpixel count with the temporally coherent split term instead of alpha (ablation) | BIST · SegTrackv2 (1300 superpixels) | SA-3D: 0.854 -> 0.853 | No clear effect |
| Adaptive iteration count (4, 8 or 12 by motion) in BIST instead of a fixed 20 iterations as in BASS | BIST · Runtime | wall-clock runtime: over twice as fast as BASS | Improved |
| Propagate superpixels via shift-and-fill from optical flow as initialization | BIST · SegTrackv2/DAVIS | temporal coherence / runtime | Improved |
| Use informative priors in boundary updates | BIST · superpixel quality | superpixel quality | No clear effect |
| Use BIST instead of TSP for temporal extent | BIST · SegTrackv2 | TEX | Hurt |
Closest papers by meaning
- Bayesian Adaptive Superpixel Segmentation (ICCV 2019 · similarity 0.87)
- BUFF: Bayesian Uncertainty Guided Diffusion Probabilistic Model for Single Image Super-Resolution (AAAI 2025 · similarity 0.83)
- Superpoint Gaussian Splatting for Real-Time High-Fidelity Dynamic Scene Reconstruction (ICML 2024 · similarity 0.83)
- Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting (ICLR 2024 · similarity 0.82)
- Spacetime Gaussian Feature Splatting for Real-Time Dynamic View Synthesis (CVPR 2024 · similarity 0.82)
- Learning Invariant Inter-pixel Correlations for Superpixel Generation (AAAI 2024 · similarity 0.82)
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.
