Research map / Point clouds / Segmentation and detection
LiDAL: Inter-Frame Uncertainty Based Active Learning for 3D LiDAR Semantic Segmentation
Zeyu Hu, Xuyang Bai, Runze Zhang, Xin Wang, Guangyuan Sun, Hongbo Fu, Chiew-Lan Tai
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 |
|---|---|---|---|
| select regions instead of whole frames (ablation) | LiDAL (FD) · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 51.8 -> 52.5 | Improved |
| add NMS-style top-region selection (ablation) | FD + region-level · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 52.5 -> 55.5 | Improved |
| add inter-frame entropy (FE) selection (ablation) | FD + region-level + NMS · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 55.5 -> 56.4 | Improved |
| add FE selection instead of NMS (ablation) | FD + region-level · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 52.5 -> 56.4 | Improved |
| add pseudo-labels from self-training (ablation) | FD + region-level + FE · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 56.4 -> 57.1 | Improved |
| use FD first then FE rather than FE first then FD (ablation) | LiDAL · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 55.7 -> 57.1 | Improved |
| raise pseudo-label target from 0.0% to 1.0% (ablation) | LiDAL · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 56.4 -> 57.1 | Improved |
| raise pseudo-label target from 1.0% to 2.0% (ablation) | LiDAL · SemanticKITTI Val (SPVCNN, first AL round) | mIoU: 57.1 -> 56.4 | Hurt |
| S3 (non-duplicate replacement) rather than S1 (growing set) (ablation) | LiDAL self-training · SemanticKITTI Val (SPVCNN, 5% labeled points) | mIoU: 58.8 -> 59.5 | Improved |
| S3 rather than S2 (constant set, replace) (ablation) | LiDAL self-training · SemanticKITTI Val (SPVCNN, 5% labeled points) | mIoU: 59.0 -> 59.5 | Improved |
| LiDAL vs best random baseline RANDre | SPVCNN · SemanticKITTI Val (5% labeled points) | mIoU: 58.2 -> 59.5 | Improved |
| LiDAL vs CSET on nuScenes | SPVCNN · nuScenes Val (5% labeled points) | mIoU: 63.0 -> 68.2 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ReDAL — ReDAL: Region-Based and Diversity-Aware Active Learning for Point Cloud Semantic Segmentation (ICCV 2021) · better on 2 of 2 shared results
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
- Annotator: A Generic Active Learning Baseline for LiDAR Semantic Segmentation (NeurIPS 2023 · similarity 0.91)
- MemorySeg: Online LiDAR Semantic Segmentation with a Latent Memory (ICCV 2023 · similarity 0.89)
- LaserMix for Semi-Supervised LiDAR Semantic Segmentation (CVPR 2023 · similarity 0.89)
- Walking Your LiDOG: A Journey Through Multiple Domains for LiDAR Semantic Segmentation (ICCV 2023 · similarity 0.88)
- Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation (CVPR 2022 · similarity 0.88)
- Exploring Scene Affinity for Semi-Supervised LiDAR Semantic Segmentation (CVPR 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.
