Research map / Point clouds / Segmentation and detection
Joint Learning of 2D-3D Weakly Supervised Semantic Segmentation
Hyeokjun Kweon, Kuk-Jin Yoon
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 2D-to-3D joint loss (ablation) | 3D WSSS baseline · ScanNetV2 val | mIoU (%): 16.2 -> 43.8 | Improved |
| add 2D-to-3D joint loss (ablation) | 3D WSSS baseline · ScanNetV2 train | mIoU (%): 17.9 -> 51.3 | Improved |
| add 3D-to-2D joint loss (ablation) | 2D WSSS baseline · ScanNetV2 val | mIoU (%): 33.3 -> 39.2 | Improved |
| add 3D-to-2D joint loss (ablation) | 2D WSSS baseline · ScanNetV2 train | mIoU (%): 38.0 -> 47.8 | Improved |
| add confidence-based scoring to 2D-to-3D loss (ablation) | 3D joint · ScanNetV2 val | mIoU (%): 43.8 -> 45.6 | Improved |
| add confidence-based scoring to 3D-to-2D loss (ablation) | 2D joint · ScanNetV2 val | mIoU (%): 39.2 -> 41.6 | Improved |
| use both bidirectional joint losses instead of one direction (ablation) | 3D joint · ScanNetV2 val | mIoU (%): 45.6 -> 49.6 | Improved |
| use both bidirectional joint losses instead of one direction (ablation) | 2D joint · ScanNetV2 val | mIoU (%): 41.6 -> 44.5 | Improved |
| replace 2D-to-3D loss with image-level classification loss on point cloud (ablation) | 3D WSSS · ScanNetV2 val | mIoU (%): 43.8 -> 39.8 | Hurt |
| use image-level labels with 3D classifier versus PCAM subcloud labels | 3D WSSS · ScanNetV2 val | mIoU (%): 38.1 -> 39.8 | No clear effect |
| add 3D-to-2D loss on unseen dataset (ablation) | 2D WSSS · S3DIS val | mIoU (%): 36.9 -> 43.3 | Improved |
| add full bidirectional joint loss on unseen dataset (ablation) | 2D WSSS · S3DIS val | mIoU (%): 36.9 -> 49.5 | Improved |
Closest papers by meaning
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.90)
- From SAM to CAMs: Exploring Segment Anything Model for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.90)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.90)
- Separate and Conquer: Decoupling Co-occurrence via Decomposition and Representation for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Less Is More: Label Recommendation for Weakly Supervised Point Cloud Semantic Segmentation (AAAI 2024 · similarity 0.88)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (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.
