Research map / Detection and segmentation / 3D detection and driving
ST3D: Self-Training for Unsupervised Domain Adaptation on 3D Object Detection
Check 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 random object scaling during source pre-training (ablation) | ST3D · Waymo → KITTI | AP3D: 34.76 -> 61.83 | Helped |
| Add triplet box partition to naive self-training (ablation) | ST (w/ SN) · Waymo → KITTI | AP3D: 65.88 -> 67.39 | Helped |
| Add the quality-aware IoU criterion (ablation) | ST (w/ SN) + Triplet · Waymo → KITTI | AP3D: 67.39 -> 70.64 | Helped |
| Add consistency memory ensemble-and-voting (ablation) | ST (w/ SN) + Triplet + QAC · Waymo → KITTI | AP3D: 70.64 -> 72.52 | Helped |
| Add curriculum data augmentation (ablation) | ST (w/ SN) + Triplet + QAC + MEV-C · Waymo → KITTI | AP3D: 72.52 -> 73.37 | Helped |
| Remove memory voting (ablation) | ST3D with consistency memory ensemble · Waymo → KITTI | AP3D: 73.37 -> 70.86 | Hurt |
| Use weighted-average instead of max-score box merging (ablation) | ST3D with consistency memory ensemble · Waymo → KITTI | AP3D: 73.37 -> 72.07 | Hurt |
| Replace curriculum augmentation intensity with strong augmentation (ablation) | ST3D · Waymo → KITTI | AP3D: 72.52 -> 72.23 | Hurt |
| Remove world-level and object-level self-training augmentation (ablation) | ST3D · Waymo → KITTI | AP3D: 72.52 -> 66.73 | Hurt |
| Adapt using ST3D rather than source-only evaluation | SECOND-IoU · Waymo → KITTI | AP3D: 27.48 -> 61.83 | Helped |
| Adapt using ST3D rather than SN | SECOND-IoU · nuScenes → KITTI | AP3D: 21.23 -> 54.13 | Helped |
| Combine ST3D with SN size statistics | ST3D · Waymo → KITTI | AP3D: 61.83 -> 73.37 | Helped |
Closest papers by meaning
- Pseudo Label Refinery for Unsupervised Domain Adaptation on Cross-dataset 3D Object Detection (CVPR 2024 · similarity 0.91)
- One for All: Multi-Domain Joint Training for Point Cloud Based 3D Object Detection (NeurIPS 2024 · similarity 0.89)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · similarity 0.89)
- SP3D: Boosting Sparsely-Supervised 3D Object Detection via Accurate Cross-Modal Semantic Prompts (CVPR 2025 · similarity 0.88)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.88)
- UNION: Unsupervised 3D Object Detection using Object Appearance-based Pseudo-Classes (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.