Research map / Detection and segmentation / 3D detection and driving
LidarMultiNet: Towards a Unified Multi-Task Network for LiDAR Perception
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the past two LiDAR frames (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 69.90 -> 70.49 | Helped |
| Add Global Context Pooling (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 70.49 -> 71.43 | Helped |
| Add the auxiliary BEV segmentation loss (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 71.43 -> 71.58 | Helped |
| Add the 3D object detection loss (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 71.58 -> 72.06 | Helped |
| Add second-stage refinement (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 72.06 -> 72.40 | Helped |
| Add test-time augmentation (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 72.40 -> 73.05 | Helped |
| Add model ensembling (ablation) | LidarMultiNet · Waymo 3D semantic segmentation validation | mIoU: 73.05 -> 73.78 | Helped |
| Train detection and segmentation jointly (ablation) | LidarMultiNet · Waymo validation | mIoU: 71.58 -> 71.93 | Helped |
| Train detection and segmentation jointly (ablation) | LidarMultiNet · Waymo validation | L2 mAPH: 75.01 -> 75.15 | Helped |
| Add second-stage segmentation refinement (ablation) | LidarMultiNet · nuScenes panoptic segmentation validation | PQ: 81.2 -> 81.8 | Helped |
| Add second-stage segmentation refinement (ablation) | LidarMultiNet · nuScenes semantic segmentation validation | mIoU: 81.7 -> 82.0 | Helped |
| Use a unified multi-task model instead of separate single-task models | LidarMultiNet · nuScenes test | mIoU: 80.4 -> 81.4 | Helped |
Closest papers by meaning
- LinK: Linear Kernel for LiDAR-Based 3D Perception (CVPR 2023 · similarity 0.90)
- LiDAR-Based Panoptic Segmentation via Dynamic Shifting Network (CVPR 2021 · similarity 0.89)
- Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation (CVPR 2021 · similarity 0.89)
- Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation (CVPR 2022 · similarity 0.89)
- Multi-Space Alignments Towards Universal LiDAR Segmentation (CVPR 2024 · similarity 0.88)
- Is Your LiDAR Placement Optimized for 3D Scene Understanding? (NeurIPS 2024 · 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.