Research map / Video / Video segmentation and tracking
SVQNet: Sparse Voxel-Adjacent Query Network for 4D Spatio-Temporal LiDAR Semantic Segmentation
Xuechao Chen, Shuangjie Xu, Xiaoyi Zou, Tongyi Cao, Dit-Yan Yeung, Lu Fang
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 Context Activator to a naive frame-stacking backbone (ablation) | DRINet multi-scan backbone · SemanticKITTI val seq 08 | mIoU (%): 52.8 -> 53.9 | Improved |
| add Sparse Voxel-Adjacent Query module (ablation) | backbone with CA · SemanticKITTI val seq 08 | mIoU (%): 53.9 -> 54.8 | Improved |
| add Temporal Feature Inheritance (ablation) | backbone with CA and SVAQ · SemanticKITTI val seq 08 | mIoU (%): 54.8 -> 55.3 | Improved |
| add Context Activator to a stacking backbone (ablation) | DRINet multi-scan backbone · SemanticKITTI val seq 08 | latency: 125 ms -> 61 ms | Improved |
| keep 45k rather than 40k activated historical voxels (ablation) | Context Activator · SemanticKITTI val seq 08 | mIoU (%): 55.3 -> 55.2 | Hurt |
| reduce activated historical voxels from 40k to 30k (ablation) | Context Activator · SemanticKITTI val seq 08 | mIoU (%): 55.3 -> 54.9 | Hurt |
| replace frame stacking with SVQNet modules on the same input | DRINet backbone · SemanticKITTI val seq 08, beyond 36m | mIoU (%): 36.1 -> 42.3 | Improved |
| plug CA, SVAQ and TFI into SPVCNN instead of stacking | SPVCNN · SemanticKITTI val multi-scan | mIoU (%): 50.04 -> 52.36 | Improved |
| use SVQNet instead of Backbone+ stacking | DRINet backbone · SemanticKITTI val seq 08 | GPU memory (MB): 8893 -> 6854 | Improved |
Closest papers by meaning
- PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation (CVPR 2020 · similarity 0.91)
- Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation (CVPR 2022 · similarity 0.90)
- SFPNet: Sparse Focal Point Network for Semantic Segmentation on General LiDAR Point Clouds (ECCV 2024 · similarity 0.90)
- SDNet: LiDAR Semantic Scene Completion with Sparse-Dense Fusion and Input-Aware Label Refinement (AAAI 2026 · similarity 0.90)
- 4D Panoptic LiDAR Segmentation (CVPR 2021 · similarity 0.89)
- (AF)2-S3Net: Attentive Feature Fusion With Adaptive Feature Selection for Sparse Semantic Segmentation Network (CVPR 2021 · 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.
