Research map / Detection and segmentation / 3D detection and driving
Sequential Multi-View Fusion Network for Fast LiDAR Point Motion Estimation
Gang Zhang, Xiaoyan Li, Zhenhua 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 |
|---|---|---|---|
| add BEV branch with multi-scan input to RV-only model (ablation) | SMVF · SemanticKITTI MOS val | moving IoU: 63.8 -> 73.5 | Improved |
| add single-scan RV branch to BEV-only model (ablation) | SMVF · SemanticKITTI MOS val | moving IoU: 73.5 -> 76.9 | Improved |
| feed multiple scans instead of a single scan to the RV branch (ablation) | SMVF · SemanticKITTI MOS val | moving IoU: 76.9 -> 76.7 | No clear effect |
| increase history scans from k=1 to k=2 (ablation) | SMVF · SemanticKITTI MOS val | moving IoU: 75.1 -> 76.9 | Improved |
| increase history scans from k=2 to k=4 (ablation) | SMVF · SemanticKITTI MOS val | moving IoU: 76.9 -> 76.9 | No clear effect |
| add BEV branch to RV-only model (ablation) | SMVF · Waymo val Vehicle | mean L2 error (m/s): 0.70 -> 0.17 | Improved |
| use multiple scans in RV branch instead of single scan (ablation) | SMVF · Waymo val Vehicle | mean L2 error (m/s): 0.17 -> 0.17 | Hurt |
| add SICP augmentation | SMVF · Waymo val Cyclist | mean L2 error (m/s): 0.38 -> 0.35 | Improved |
| add SICP augmentation | MotionNet* · Waymo val Cyclist | mean L2 error (m/s): 0.35 -> 0.31 | Improved |
| add SICP when pedestrian category treated as stationary | SMVF · Waymo generalization | pedestrian mean error (m/s): 0.79 -> 0.70 | Improved |
| add SICP when pedestrian category is ignored | SMVF · Waymo generalization | pedestrian mean error (m/s): 0.57 -> 0.36 | Improved |
| label ablated cyclist as ignored instead of stationary (ablation) | FastFlow3D · Waymo generalization | cyclist mean error (m/s): 1.13 -> 0.83 | Improved |
Closest papers by meaning
- MotionNet: Joint Perception and Motion Prediction for Autonomous Driving Based on Bird's Eye View Maps (CVPR 2020 · similarity 0.89)
- PC-BEV: An Efficient Polar-Cartesian BEV Fusion Framework for LiDAR Semantic Segmentation (AAAI 2025 · similarity 0.88)
- Exploiting the Benefits of Temporal Information in the Realm of LiDAR Panoptic Segmentation (TPAMI 2025 · similarity 0.88)
- SynFlow: Scaling Up LiDAR Scene Flow Estimation with Synthetic Data (ECCV 2026 · similarity 0.88)
- ICP-Flow: LiDAR Scene Flow Estimation with ICP (CVPR 2024 · similarity 0.88)
- Effective Multi-sensor Conditioning for Street-view Novel-view Synthesis (NeurIPS 2026 · 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.
