Research map / Point clouds / Segmentation and detection
TCC-Det: Temporarily consistent cues for weakly-supervised 3D detection
Jan Skvrna, Lukáš Neumann
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 Appearance Mask Loss to the training objective (ablation) | TCC-Det · KITTI val 3D AP Easy @0.7 | 3D AP: 78.42 -> 85.35 | Improved |
| add Template Fitting Loss alone (without AML) (ablation) | TCC-Det · KITTI val 3D AP Easy @0.7 | 3D AP: 78.42 -> 76.42 (Easy), 64.77 -> 66.01 (Moderate) | Mixed |
| combine TFL and AML (ablation) | TCC-Det · KITTI val 3D AP Hard @0.7 | 3D AP: 63.64 -> 73.74 | Improved |
| use Template Fitting Loss instead of Chamfer distance loss (ablation) | TCC-Det · KITTI val 3D AP Moderate @0.7 | 3D AP: 53.93 -> 75.33 | Improved |
| use Template Fitting Loss instead of McCraith loss (ablation) | TCC-Det · KITTI val 3D AP Moderate @0.7 | 3D AP: 49.76 -> 75.33 | Improved |
| fine-tune with 25% human labels (ablation) | TCC-Det · KITTI val 3D AP Moderate @0.7 | 3D AP: 75.33 -> 78.98 | Improved |
| increase human labels from 25% to 100% for fine-tuning (ablation) | TCC-Det · KITTI val 3D AP Hard @0.7 | 3D AP: 78.13 -> 78.77 | No clear effect |
| train on Waymo without 3D labels instead of fully supervised | TCC-Det · Waymo Open val Level 1 3D AP @0.7 | 3D AP: 77.37 -> 17.44 | Hurt |
| aggregate LiDAR points across frames (temporal consistency) (ablation) | TCC-Det · KITTI | 3D AP | Improved |
Closest papers by meaning
- Weakly Supervised 3D Object Detection from Lidar Point Cloud (ECCV 2020 · similarity 0.93)
- Towards A Weakly Supervised Framework for 3D Point Cloud Object Detection and Annotation (TPAMI 2021 · similarity 0.93)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.91)
- AS-Det: Active Sampling for Adaptive 3D Object Detection in Point Clouds (AAAI 2025 · similarity 0.89)
- STONE: A Submodular Optimization Framework for Active 3D Object Detection (NeurIPS 2024 · similarity 0.89)
- A-Teacher: Asymmetric Network for 3D Semi-Supervised 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.
