Research map / Point clouds / Point cloud classification
Learning Scene Dynamics from Point Cloud Sequences
Pan Li He, Patrick Emami, Sanjay Ranka, Anand Rangarajan
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 |
|---|---|---|---|
| retrain pair-wise model on multi-step SFT3D sequences instead of pair-wise FT3D (ablation) | PointPWC-Net · SFT3D validation (SSFE) | EPE3D: 0.190 -> 0.115 | Improved |
| retrain pair-wise model on SFT3D sequences instead of FT3D (ablation) | FlowNet3D · SFT3D validation (SSFE) | EPE3D: 0.200 -> 0.136 | Improved |
| retrain pair-wise model on SFT3D sequences instead of FT3D (ablation) | FLOT · SFT3D validation (SSFE) | EPE3D: 0.173 -> 0.159 | Improved |
| add recurrent cost volume over the sequence (SPCM-Net vs PointPWC-Net + SFT3D) | PointPWC-Net · SFT3D validation (SSFE) | EPE3D: 0.115 -> 0.108 | Improved |
| add recurrent cost volume (SPCM-Net vs best baselines) | PointPWC-Net / FlowNet3D · SFT3D test (SSFE) | EPE3D / Acc3DS: EPE3D 0.150 (FlowNet3D) vs 0.157; Acc3DS 0.320 -> 0.380 | Mixed |
| add recurrent processing (SPCM-Net vs PointPWC-Net) | PointPWC-Net · VKS test (SSFE) | EPE3D: 0.0458 -> 0.0454 | Improved |
| add SSFE pretraining before SPF finetuning (ablation) | SPCM-Net · SFT3D test (SPF) | ADE: 0.6286 -> 0.3978 | Improved |
| add SSFE pretraining before SPF finetuning (ablation) | SPCM-Net · SFT3D validation (SPF) | ADE: 0.5069 -> 0.2488 | Improved |
| replace PointRNN with SPCM-Net for forecasting | PointRNN · VKS (SPF) | ADE: 0.2856 -> 0.2768 | No clear effect |
| replace PointRNN with SPCM-Net for self-supervised forecasting | PointRNN · SAG Argoverse (self-supervised SPF) | CD: 1.2322 -> 1.3453 | Mixed |
| replace PointRNN with SPCM-Net for forecasting | PointRNN · nuScenes (SPF) | CD: 0.9750 -> 0.6339 | Improved |
| add one more preceding frame (3 to 4) (ablation) | SPCM-Net · KITTI scene flow (mean EPE) | EPE (mean): 0.229 -> 0.194 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PointPWC-Net — PointPWC-Net: Cost Volume on Point Clouds for (Self-)Supervised Scene Flow Estimation (ECCV 2020) · better on 1 of 1 shared result
- FlowNet3D — FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019) · better on 1 of 1 shared result
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
- SynFlow: Scaling Up LiDAR Scene Flow Estimation with Synthetic Data (ECCV 2026 · similarity 0.91)
- GenFlow3D: Generative Scene Flow Estimation and Prediction on Point Cloud Sequences (ICCV 2025 · similarity 0.91)
- MESSENGER: Memory-Enhanced Sequential Scene Flow Estimation via Autoregressive Next-Frame Forecasting (NeurIPS 2026 · similarity 0.90)
- ZeroFlow: Scalable Scene Flow via Distillation (ICLR 2024 · similarity 0.89)
- ICP-Flow: LiDAR Scene Flow Estimation with ICP (CVPR 2024 · similarity 0.89)
- FlowMamba: Learning Point Cloud Scene Flow with Global Motion Propagation (AAAI 2025 · similarity 0.89)
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.
