Research map / Label-efficient and robust learning / Test-time adaptation
4D Point Cloud Segmentation via Active Test-Time Adaptation
Mingrong Gong, Chaoqi Chen, Luyao Tang, Yuxi Wang, Sergio Escalera
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 target prototype update with oracle labels (ablation) | ATTA-4DSeg · SynLiDAR -> SemanticKITTI | mIoU improvement: 17.12 -> 18.87 | Improved |
| add voxel uncertainty to point uncertainty (ablation) | ATTA-4DSeg · SynLiDAR -> SemanticKITTI | mIoU improvement: 13.90 -> 18.87 | Improved |
| add point uncertainty to voxel uncertainty (ablation) | ATTA-4DSeg · SynLiDAR -> SemanticKITTI | mIoU improvement: 15.89 -> 18.87 | Improved |
| allocate budget only to Vehicle class instead of all 7 classes (CIBA coverage) (ablation) | ATTA-4DSeg · SynLiDAR -> SemanticKITTI | mIoU improvement: -12.1 avg -> +18.9 | Hurt |
| replace pseudo-label TTA (HGL) with active TTA | ATTA-4DSeg · SynLiDAR -> SemanticKITTI | mIoU improvement: +6.72 -> +18.87 | Improved |
| replace random active selection with ATTA-4DSeg selection | ATTA-4DSeg · Synth4D -> nuScenes | mIoU improvement: +2.88 -> +3.62 | Improved |
| use ATTA-4DSeg instead of HGL | ATTA-4DSeg · Synth4D -> SemanticKITTI | inference time (s): 2.76 -> 1.21 | Improved |
| use entropy-based TTA (TENT) or ProDA/SHOT on LiDAR | TTA baselines · SynLiDAR -> SemanticKITTI | mIoU improvement: ProDA -38.60, SHOT -36.89, TENT +1.15 | Mixed |
| increase annotation budget to 5% (ablation) | ATTA-4DSeg · Synth4D -> SemanticKITTI | mIoU: 6.2 (0.1%) -> 22.3 (5%) | 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.
- TENT — Tent: Fully Test-Time Adaptation by Entropy Minimization (ICLR 2021) · 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
- Annotator: A Generic Active Learning Baseline for LiDAR Semantic Segmentation (NeurIPS 2023 · similarity 0.90)
- D^3CTTA: Domain-Dependent Decorrelation for Continual Test-Time Adaption of 3D LiDAR Segmentation (CVPR 2025 · similarity 0.90)
- Learning Temporal Variations for 4D Point Cloud Segmentation (IJCV 2024 · similarity 0.89)
- Less Is More: Reducing Task and Model Complexity for 3D Point Cloud Semantic Segmentation (CVPR 2023 · similarity 0.88)
- Point-Cache: Test-time Dynamic and Hierarchical Cache for Robust and Generalizable Point Cloud Analysis (CVPR 2025 · similarity 0.88)
- AS-Det: Active Sampling for Adaptive 3D Object Detection in Point Clouds (AAAI 2025 · similarity 0.86)
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.
