Research map / 3D vision / Depth and stereo
Superpixel Soup: Monocular Dense 3D Reconstruction of a Complex Dynamic Scene
Suryansh Kumar, Yuchao Dai, Hongdong 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 |
|---|---|---|---|
| add projection energy to ARAP (ablation) | Superpixel Soup · MPI Sintel alley_1 | mean relative error: 0.2248 -> 0.2022 | Improved |
| add 3D continuity energy (ablation) | Superpixel Soup · MPI Sintel alley_1 | mean relative error: 0.2022 -> 0.1697 | Improved |
| add normal orientation energy (ablation) | Superpixel Soup · MPI Sintel alley_1 | mean relative error: 0.1697 -> 0.1606 | Improved |
| add normal orientation energy (ablation) | Superpixel Soup · MPI Sintel ambush_4 | mean relative error: 0.1701 -> 0.1676 | No clear effect |
| add 3D continuity energy (ablation) | Superpixel Soup · MPI Sintel sleeping_1 | mean relative error: 0.2026 -> 0.1912 | Improved |
| add projection energy to ARAP (ablation) | Superpixel Soup · MPI Sintel mountain_1 | mean relative error: 0.2127 -> 0.1923 | Improved |
| use SLIC superpixels instead of regular grid (ablation) | Superpixel Soup · single frame | mean relative error: 0.1442 -> 0.0912 | Improved |
| use two-frame Superpixel Soup instead of multi-frame BMM | Superpixel Soup · kinect_paper | mean relative error: 0.0322 -> 0.0472 | Hurt |
| use Superpixel Soup instead of DMDE | Superpixel Soup · KITTI | mean relative error: 0.148 -> 0.1254 | Improved |
| use Superpixel Soup instead of DMDE | Superpixel Soup · MPI Sintel | mean relative error: 0.297 -> 0.1643 | Improved |
| use more than 1000 superpixels (ablation) | Superpixel Soup · depth error vs superpixel count | mean relative error | No clear effect |
Closest papers by meaning
- Monocular Dense 3D Reconstruction of a Complex Dynamic Scene From Two Perspective Frames (ICCV 2017 · similarity 0.96)
- Planar Structure-from-Motion with Affine Camera Models: Closed-Form Solutions, Ambiguities and Degeneracy Analysis (TPAMI 2016 · similarity 0.88)
- Real-Time Large-Scale Dense 3D Reconstruction with Loop Closure (ECCV 2016 · similarity 0.88)
- Motion Decoupled 3D Gaussian Splatting for Dynamic Object Representation (AAAI 2025 · similarity 0.86)
- Dense-SfM: Structure from Motion with Dense Consistent Matching (CVPR 2025 · similarity 0.86)
- Enhancing 3D Reconstruction for Dynamic Scenes (NeurIPS 2025 · similarity 0.86)
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.
