Research map / 3D vision / 3D generation
The Structure-Equivalent Prior: Unifying Temporal Dynamics and 3D Evolution in 4D Latent Space
Jingyuan Gao, Tianyu Shen, Ruosen Hao, Te Guo, Zhiwei Li, Kunfeng Wang
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 |
|---|---|---|---|
| replace concatenation of feature planes with plane-wise spatio-temporal fusion (ablation) | SEP-4D · CarlaSC (16 frames) | mIoU (%): 63.71 -> 76.41 | Improved |
| replace averaging of planes with plane-wise fusion (ablation) | SEP-4D · CarlaSC (16 frames) | mIoU (%): 75.39 -> 76.41 | Improved |
| replace dynamic weighting with plane-wise fusion (ablation) | SEP-4D · CarlaSC (16 frames) | mIoU (%): 75.61 -> 76.41 | Improved |
| replace conv fusion with plane-wise fusion (ablation) | SEP-4D · CarlaSC (16 frames) | mIoU (%): 74.78 -> 76.41 | Improved |
| use planar factorization encoder instead of Conv3D + SE block on full voxels (ablation) | SEP-4D · CarlaSC (16 frames) | mIoU (%): 46.74 -> 76.41 | Improved |
| use SEP-4D instead of DynamicCity | 4D occupancy reconstruction · CarlaSC (4 frames) | mIoU (%): 76.25 -> 91.68 | Improved |
| use SEP-4D instead of DynamicCity | 4D occupancy reconstruction · CarlaSC (64 frames) | mIoU (%): 47.17 -> 56.80 | Improved |
| use SEP-4D instead of DynamicCity | 4D occupancy reconstruction · CarlaSC (64 frames) | training time (s): 11156 s -> 13208 s | Hurt |
| use SEP-4D instead of DynamicCity | 4D occupancy reconstruction · CarlaSC (4 frames) | training time (s): 1912 s -> 1474 s | Improved |
| use SEP-4D instead of DynamicCity | 4D occupancy reconstruction · CarlaSC Pedestrian class (16 frames) | IoU (%): 49.844 -> 44.034 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DynamicCity — DynamicCity: Large-Scale 4D Occupancy Generation from Dynamic Scenes (ICLR 2025) · better on 6 of 8 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
- Tensor4D: Efficient Neural 4D Decomposition for High-Fidelity Dynamic Reconstruction and Rendering (CVPR 2023 · similarity 0.91)
- Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting (ICLR 2024 · similarity 0.90)
- MoBa-GS: Learning a Spatially-Varying Motion Basis over a Dynamic Canonical Space for 4D Reconstruction (ECCV 2026 · similarity 0.89)
- Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models (NeurIPS 2024 · similarity 0.89)
- 4DWeaver: Bridging Reconstruction and Generation Via Compact Autoregressive Priors (TPAMI 2026 · similarity 0.89)
- Motion4D: Learning 3D-Consistent Motion and Semantics for 4D Scene Understanding (NeurIPS 2025 · similarity 0.88)
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.
