Research map / Detection and segmentation / 3D detection and driving
A Simple Vision Transformer for Weakly Semi-supervised 3D Object Detection
Dingyuan Zhang, Dingkang Liang, Zhikang Zou, Jingyu Li, Xiaoqing Ye, Zhe Liu, Xiao Tan, Xiang Bai
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 pseudo labels from ViT-WSS3D (2% full + 98% point-labeled) to student training | PointPillars · KITTI val 2% full | overall mAP40 moderate: 24.4 -> 48.3 | Improved |
| use point-annotation pseudo labels instead of 3DIoUMatch semi-supervised | PVRCNN · KITTI val 5% | overall mAP40: 60.6 -> 63.6 | Improved |
| use point-annotation pseudo labels instead of 3DIoUMatch | VoteNet · SUN RGB-D 5% full | mAP@0.25: 40.0 -> 52.4 | Improved |
| use 1024 scene tokens instead of 2048 (ablation) | ViT-WSS3D · KITTI val 5% | overall mAP40: 67.2 -> 61.9 (PointRCNN) | Hurt |
| increase scene tokens from 2048 to 3072 (ablation) | ViT-WSS3D · KITTI val 5% | overall mAP40: 67.2 -> 67.6 (PointRCNN) | No clear effect |
| increase random disturbance on point annotations to R=0.15 (ablation) | ViT-WSS3D · KITTI val 5% | overall mAP40: 67.9 -> 66.8 (PointRCNN) | Hurt |
| add 2D ViT pre-trained weights (DeiT) to the teacher (ablation) | ViT-WSS3D ViT-S · KITTI val 5% | overall mAP40: 65.5 -> 67.2 (PointRCNN) | Improved |
| use MAE instead of DeiT pre-training for ViT-B teacher (ablation) | ViT-WSS3D ViT-B · KITTI val 10% | overall mAP40: 67.1 -> 69.7 (PointRCNN) | Improved |
Closest papers by meaning
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.91)
- Leveraging Temporal Cues for Semi-Supervised Multi-View 3D Object Detection (CVPR 2025 · similarity 0.89)
- WeakPCSOD: Overcoming the Bias of Box Annotations for Weakly Supervised Point Cloud Salient Object Detection (AAAI 2024 · similarity 0.89)
- Semi-supervised 3D Object Detection with PatchTeacher and PillarMix (AAAI 2024 · similarity 0.89)
- VSRD: Instance-Aware Volumetric Silhouette Rendering for Weakly Supervised 3D Object Detection (CVPR 2024 · similarity 0.89)
- PTT: Point-Trajectory Transformer for Efficient Temporal 3D Object Detection (CVPR 2024 · similarity 0.89)
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.
