Research map / Point clouds / Segmentation and detection
Weakly Supervised Monocular 3D Object Detection Using Multi-View Projection and Direction Consistency
Runzhou Tao, Wencheng Han, Zhongying Qiu, Cheng-Zhong Xu, Jianbing Shen
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 multi-view consistency loss and direction loss on top of projection loss (ablation) | DD3D with WeakMono3D · KITTI val car | AP_3D Easy: 0.00 -> 49.37 (full) | Improved |
| remove direction consistency loss (keep projection + multi-view) (ablation) | WeakMono3D · KITTI val car | AP_3D Easy: 49.37 -> 11.78 when removed | Hurt |
| add multi-view consistency loss to projection loss (ablation) | WeakMono3D · KITTI val car | AP_3D Easy: 0.00 -> 11.78 | Improved |
| use multi-camera pairs instead of single-camera video sequences (ablation) | WeakMono3D · KITTI val car | AP_3D Easy: 36.68 -> 49.37 | Improved |
| pre-train with WeakMono3D then fine-tune with 1/3 3D labels (ablation) | DD3D · KITTI val car | AP_3D Easy: 58.08 -> 61.02 | Improved |
| pre-train with WeakMono3D then fine-tune with varying small label ratios (1/20 to 1/5) (ablation) | DD3D · KITTI val car | AP_3D Easy: 1/20: 53.61; 1/10: 57.21; 1/5: 59.12 vs DD3D 58.08 | Improved |
| weakly supervised 2D-label training vs WeakM3D (LiDAR-based weak) | WeakMono3D · KITTI test car IoU 0.7 | AP_3D Easy: 5.03 -> 6.98 | Improved |
| weakly supervised 2D-label training on pedestrians/cyclists vs fully supervised DD3D | WeakMono3D · KITTI test pedestrian/cyclist | AP_3D Easy: pedestrian 13.91 (DD3D) vs 4.62 (ours) | Hurt |
Closest papers by meaning
- Weakly Supervised Monocular 3D Detection with a Single-View Image (CVPR 2024 · similarity 0.90)
- VSRD: Instance-Aware Volumetric Silhouette Rendering for Weakly Supervised 3D Object Detection (CVPR 2024 · similarity 0.89)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.89)
- WeakPCSOD: Overcoming the Bias of Box Annotations for Weakly Supervised Point Cloud Salient Object Detection (AAAI 2024 · similarity 0.88)
- LabelAny3D: Label Any Object 3D in the Wild (NeurIPS 2025 · similarity 0.87)
- FD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection (AAAI 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.
