Research map / Point clouds / Segmentation and detection
SSDA3D: Semi-supervised Domain Adaptation for 3D Object Detection from Point Cloud
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 |
|---|---|---|---|
| Use SSDA3D instead of ST3D-SSDA | CenterPoint on Waymo→nuScenes · 5% target labels | AP: 63.9 -> 76.2 | Helped |
| Use SSDA3D instead of the fully supervised Oracle | CenterPoint on Waymo→nuScenes · 10% target labels | AP: 78.4 -> 78.8 | Helped |
| Add source-supervised training without inter-domain CutMix (ablation) | Target-supervised CenterPoint · Waymo→nuScenes, 5% target labels | AP: 61.0 -> 57.7 | Hurt |
| Add inter-domain Point-CutMix to source- and target-supervised training (ablation) | Co-training CenterPoint · Waymo→nuScenes, 5% target labels | AP: 57.7 -> 74.5 | Helped |
| Add inter-domain Point-CutMix to source- and target-supervised training (ablation) | Co-training CenterPoint · Waymo→nuScenes, 5% target labels | NDS: 58.0 -> 67.8 | Helped |
| Add naive SSL with real- and pseudo-labeled target samples (ablation) | Inter-domain CutMix model · Waymo→nuScenes, 5% target labels | AP: 74.5 -> 73.8 | Hurt |
| Add intra-domain Point-MixUp to naive SSL (ablation) | Inter-domain CutMix model · Waymo→nuScenes, 5% target labels | AP: 73.8 -> 76.2 | Helped |
| Add intra-domain Point-MixUp to naive SSL (ablation) | Inter-domain CutMix model · Waymo→nuScenes, 5% target labels | NDS: 67.7 -> 68.8 | Helped |
| Omit source-domain supervision from the training setup (ablation) | Target-only SSL + Point-MixUp · Waymo→nuScenes, 5% target labels | AP: 76.2 -> 65.4 | Hurt |
| Increase inter-domain Point-CutMix probability up to 0.5 (ablation) | SSDA3D · Waymo→nuScenes, 5% target labels | AP: null | Helped |
| Increase intra-domain Point-MixUp probability up to 0.5 (ablation) | SSDA3D · Waymo→nuScenes, 5% target labels | AP: null | Helped |
| Raise Point-MixUp probability above 0.5 (ablation) | SSDA3D · Waymo→nuScenes, 5% target labels | AP: null | Hurt |
Closest papers by meaning
- CMDA: Cross-Modal and Domain Adversarial Adaptation for LiDAR-Based 3D Object Detection (AAAI 2024 · similarity 0.90)
- One for All: Multi-Domain Joint Training for Point Cloud Based 3D Object Detection (NeurIPS 2024 · similarity 0.89)
- An Empirical Study of the Generalization Ability of Lidar 3D Object Detectors to Unseen Domains (CVPR 2024 · similarity 0.89)
- PSA-SSL: Pose and Size-aware Self-Supervised Learning on LiDAR Point Clouds (CVPR 2025 · similarity 0.89)
- GBlobs: Explicit Local Structure via Gaussian Blobs for Improved Cross-Domain LiDAR-based 3D Object Detection (CVPR 2025 · similarity 0.89)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.88)
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.