Research map / Graph learning / Knowledge and temporal graphs
4D Panoptic Scene Graph Generation
Jingkang Yang, Jun CEN, Wenxuan Peng, Shuai Liu, Fangzhou Hong, Xiangtai Li, Kaiyang Zhou, Qifeng Chen, Ziwei Liu
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 |
|---|---|---|---|
| remove temporal encoder (ablation) | PSG4DFormer (RGB-D) · PSG4D-GTA | R@20: 6.68 -> 4.43 | Hurt |
| remove temporal encoder (ablation) | PSG4DFormer (RGB-D) · PSG4D-GTA | mR@20: 3.31 -> 1.34 | Hurt |
| remove temporal encoder (ablation) | PSG4DFormer (RGB-D) · PSG4D-HOI | R@20: 5.62 -> 4.44 | Hurt |
| remove temporal encoder (ablation) | PSG4DFormer (point cloud) · PSG4D-GTA | R@20: 4.33 -> 2.25 | Hurt |
| remove temporal encoder (ablation) | PSG4DFormer (point cloud) · PSG4D-HOI | R@100: 6.76 -> 4.18 | Hurt |
| remove depth branch (RGB encoder and SA-Gate) (ablation) | PSG4DFormer (RGB-D) · PSG4D-GTA | R@20: 6.68 -> 4.40 | Hurt |
| remove depth branch (RGB encoder and SA-Gate) (ablation) | PSG4DFormer (RGB-D) · PSG4D-GTA | mR@100: 4.02 -> 2.27 | Hurt |
| remove depth branch (RGB encoder and SA-Gate) (ablation) | PSG4DFormer (RGB-D) · PSG4D-HOI | R@20: 5.62 -> 5.49 | No clear effect |
| remove depth branch (RGB encoder and SA-Gate) (ablation) | PSG4DFormer (RGB-D) · PSG4D-HOI | mR@100: 4.97 -> 4.21 | Hurt |
| use RGB-D input instead of point cloud input | PSG4DFormer · PSG4D-GTA | R@20: 4.33 -> 6.68 | Improved |
| use RGB-D input instead of point cloud input | PSG4DFormer · PSG4D-HOI | R@20: 5.36 -> 5.62 | No clear effect |
| use PSG4DFormer instead of 3DSGG baseline | RGB-D input · PSG4D-GTA | R@20: 2.29 -> 6.68 | Improved |
Closest papers by meaning
- Motion-aware Contrastive Learning for Temporal Panoptic Scene Graph Generation (AAAI 2025 · similarity 0.87)
- Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting (ICLR 2024 · similarity 0.87)
- Motion4D: Learning 3D-Consistent Motion and Semantics for 4D Scene Understanding (NeurIPS 2025 · similarity 0.87)
- X4D-SceneFormer: Enhanced Scene Understanding on 4D Point Cloud Videos through Cross-Modal Knowledge Transfer (AAAI 2024 · similarity 0.86)
- IGGT4D: Streaming 4D Instance-Grounded Geometry Transformer (NeurIPS 2026 · similarity 0.86)
- Panoptic Scene Graph Generation with Semantics-Prototype Learning (AAAI 2024 · 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.
