Research map / Detection and segmentation / Weakly supervised detection and segmentation
Hierarchical Supervision and Shuffle Data Augmentation for 3D Semi-Supervised Object Detection
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 |
|---|---|---|---|
| Apply HSSDA instead of the reproduced PV-RCNN baseline | PV-RCNN · Waymo val, 1% labeled, Cyclist Level 1 | mAP: 4.5 -> 29.1 | Helped |
| Apply HSSDA instead of the PV-RCNN baseline | PV-RCNN · KITTI val, 2% labeled | Avg. AP: 54.3 -> 68.6 | Helped |
| Add high-confidence pseudo-label supervision (ablation) | Voxel-RCNN semi-supervised training · KITTI val, 2% labeled | mAP: 52.2 -> 59.7 | Helped |
| Add ambiguous-level pseudo-label supervision (ablation) | Voxel-RCNN with high-confidence supervision · KITTI val, 2% labeled | mAP: 59.7 -> 60.7 | Helped |
| Add low-confidence point removal (ablation) | Voxel-RCNN with high- and ambiguous-level supervision · KITTI val, 2% labeled | mAP: 60.7 -> 66.5 | Helped |
| Add shuffle data augmentation (ablation) | Voxel-RCNN with all hierarchical supervision · KITTI val, 2% labeled | mAP: 66.5 -> 68.6 | Helped |
| Add shuffle data augmentation (ablation) | Voxel-RCNN with all hierarchical supervision · KITTI val, 2% labeled, Car | 3D detection AP: 82.4 -> 82.0 | Hurt |
| Add shuffle data augmentation (ablation) | Voxel-RCNN with all hierarchical supervision · KITTI val, 2% labeled, Pedestrian | 3D detection AP: 56.0 -> 58.3 | Helped |
| Add shuffle data augmentation (ablation) | Voxel-RCNN with all hierarchical supervision · KITTI val, 2% labeled, Cyclist | 3D detection AP: 61.3 -> 65.7 | Helped |
| Vary shuffled patch grid to 1 × 2 (ablation) | HSSDA with hierarchical supervision · KITTI val, 2% labeled | mAP: 66.5 -> 67.9 | Helped |
| Vary shuffled patch grid from 2 × 2 to 2 × 4 (ablation) | HSSDA · KITTI val, 2% labeled | mAP: 68.6 -> 67.8 | Hurt |
| Vary shuffled patch grid from 2 × 2 to 4 × 4 (ablation) | HSSDA · KITTI val, 2% labeled | mAP: 68.6 -> 67.2 | Hurt |
Closest papers by meaning
- Semi-supervised 3D Object Detection with PatchTeacher and PillarMix (AAAI 2024 · similarity 0.90)
- Dual-Perspective Knowledge Enrichment for Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.90)
- A-Teacher: Asymmetric Network for 3D Semi-Supervised Object Detection (CVPR 2024 · similarity 0.89)
- Leveraging Temporal Cues for Semi-Supervised Multi-View 3D Object Detection (CVPR 2025 · similarity 0.88)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.87)
- Relational Matching for Weakly Semi-Supervised Oriented Object Detection (CVPR 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.