Research map / 3D vision / 3D generation
Object-Centric Representation Learning with Generative Spatial-Temporal Factorization
Li Nanbo, Muhammad Ahmed Raza, Hu Wenbin, Zhaole Sun, Robert Burns Fisher
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 |
|---|---|---|---|
| use DyMON instead of MulMON on dynamic scenes | DyMON · DRoom multi-view-dynamic | MSE obs. reconstruction: 0.011 -> 0.004 | Improved |
| use DyMON instead of MulMON on dynamic scenes | DyMON · DRoom multi-view-dynamic | mIoU obs. segmentation: 0.511 -> 0.717 | Improved |
| use DyMON instead of MulMON on dynamic scenes | DyMON · DRoom multi-view-dynamic | mIoU novel-view segmentation: 0.461 -> 0.508 | Improved |
| use DyMON instead of MulMON on dynamic scenes | DyMON · DRoom multi-view-dynamic | MSE novel-view obs.: 0.019 -> 0.021 | Hurt |
| use DyMON instead of specialized MulMON on static scenes | DyMON · DRoom multi-view-static | MSE obs. reconstruction: 0.006 -> 0.014 | Hurt |
| use DyMON instead of specialized MulMON on static scenes | DyMON · DRoom multi-view-static | mIoU obs. segmentation: 0.583 -> 0.529 | Hurt |
| use DyMON instead of GSWM on single-view dynamic scenes | DyMON · DRoom single-view-dynamic | MSE obs. reconstruction: 0.039 -> 0.014 | Improved |
| use DyMON instead of GSWM on single-view dynamic scenes | DyMON · DRoom single-view-dynamic | mIoU obs. segmentation: 0.402 -> 0.682 | Improved |
| train on faster speed-difference levels DR-Lvl.1 to 3 (ablation) | DyMON · DRoom space-time prediction | MSE | Improved |
| vary update periods of z (3,5) and v (5,6,8) and beta (ablation) | DyMON · DRoom novel-view synthesis | MSE | Mixed |
| train T-GQN on DRoom DR-Lvl.3, with or without posterior dropout | T-GQN · DRoom DR-Lvl.3 | scene representation | No clear effect |
Closest papers by meaning
- DynaVol: Unsupervised Learning for Dynamic Scenes through Object-Centric Voxelization (ICLR 2024 · similarity 0.89)
- DORSal: Diffusion for Object-centric Representations of Scenes $\textit{et al.}$ (ICLR 2024 · similarity 0.88)
- Temporally Consistent Object-Centric Learning by Contrasting Slots (CVPR 2025 · similarity 0.87)
- EmerNeRF: Emergent Spatial-Temporal Scene Decomposition via Self-Supervision (ICLR 2024 · similarity 0.87)
- OSN: Infinite Representations of Dynamic 3D Scenes from Monocular Videos (ICML 2024 · similarity 0.87)
- Dyn-O: Building Structured World Models with Object-Centric Representations (NeurIPS 2025 · similarity 0.87)
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.
