Research map / 3D vision / 3D generation
HieraScaffold: Learning Compact Hierarchical Representations for Scalable 4D LiDAR Generation
Zijie Wu, Na Zhao
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 hierarchical scaffold generation versus prior LiDAR generators | HieraScaffold · KITTI-360 generation | FRD: 1012.5 (LiDMs) -> 979.3 | Improved |
| add multi-resolution node features in scaffold (ablation) | HieraScaffold · Waymo reconstruction | Chamfer Distance: 0.3587 -> 0.3214 | Improved |
| add joint tuning of contourlet filters with frozen NF (ablation) | HieraScaffold · Waymo reconstruction | F-Score: 0.8493 -> 0.8521 | Improved |
| use scaffold representation instead of Hexplane (ablation) | HieraScaffold · Waymo reconstruction | Chamfer Distance: 0.3879 -> 0.3214 | Improved |
| use transformer NF instead of U-net NF (ablation) | HieraScaffold · Waymo reconstruction | Chamfer Distance: 0.3422 -> 0.3214 | Improved |
| use learnable neural filter instead of fixed Biorthogonal filter (ablation) | NCR · KITTI-360/Waymo generation | FRD: 1205.3 -> 1087.4 | Improved |
| add coarse2fine generation (ablation) | HieraScaffold · generation | FRD: 1126.7 -> 1087.4 | Improved |
| add directional filter bank (ablation) | HieraScaffold · generation | FRD: 1143.5 -> 1087.4 | Improved |
| generate dynamics conditioned on static (S2D) instead of jointly (S&D) (ablation) | HieraScaffold · generation | FRD: 1129.4 -> 1087.4 | Improved |
| augment 1k real data with 10k generated samples | segmentation model training · evaluation on real data | overall mIoU: 68.8 -> 76.2 | Improved |
Closest papers by meaning
- LiDAR4D: Dynamic Neural Fields for Novel Space-time View LiDAR Synthesis (CVPR 2024 · similarity 0.90)
- 3D LiDAR Mapping in Dynamic Environments using a 4D Implicit Neural Representation (CVPR 2024 · similarity 0.89)
- TREND: Unsupervised 3D Representation Learning via Temporal Forecasting for LiDAR Perception (NeurIPS 2025 · similarity 0.88)
- LiFi: LiDAR Generation from Multi-View Images via Geometric and Semantic Collaborative Guidance (NeurIPS 2026 · similarity 0.87)
- Outdoor Scene Extrapolation with Hierarchical Generative Cellular Automata (CVPR 2024 · similarity 0.87)
- STGC-NeRF: Spatial-Temporal Geometric Consistency for LiDAR Neural Radiance Fields in Dynamic Scenes (AAAI 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.
