Research map / Point clouds / Segmentation and detection
Quasi-Balanced Self-Training on Noise-Aware Synthesis of Object Point Clouds for Closing Domain Gap
Yongwei Chen, Zihao Wang, Longkun Zou, Ke Chen, Kui Jia
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 |
|---|---|---|---|
| train on stereo-rendered SpeckleNet10 instead of ModelNet10 | DGCNN · DepthScanNet10 | accuracy: 46.7 -> 64.0 | Improved |
| train on SpeckleNet10 instead of ModelNet10 | PointNet++ · DepthScanNet10 | accuracy: 48.4 -> 60.9 | Improved |
| train on SpeckleNet10 instead of ModelNet10 | RSCNN · DepthScanNet10 | accuracy: 49.7 -> 53.9 | Improved |
| train on SpeckleNet10 instead of ModelNet10 | SimpleView · DepthScanNet10 | accuracy: 54.6 -> 62.3 | Improved |
| use DepthScanNet-like noise via stereo matching instead of clean depth (add stereo rendering and matching noise) (ablation) | DGCNN · S->D | accuracy: 53.2 -> 64.0 | Improved |
| use Mesh2Point instead of Blensor simulation (ablation) | DGCNN · S->D | accuracy: 56.6 -> 64.0 | Improved |
| use Mesh2Point instead of ModelNet10 with random region dropout (ablation) | DGCNN · S->D | accuracy: 50.4 -> 64.0 | Improved |
| use Mesh2Point instead of Blensor (ablation) | SimpleView · S->D | accuracy: 65.1 -> 62.3 | Hurt |
| use Mesh2Point instead of ModelNet10-Dropout (ablation) | RSCNN · S->D | accuracy: 56.5 -> 53.9 | Hurt |
| add PCM mixup augmentation | DGCNN · M->D | accuracy: 46.7 -> 61.2 | Improved |
| add QBST on top of PCM instead of SPST | DGCNN+PCM · S->D | accuracy: 71.9 -> 72.4 | Improved |
| use QBST instead of CBST on top of PCM | DGCNN+PCM · S->D | accuracy: 62.9 -> 72.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PointDAN — PointDAN: A Multi-Scale 3D Domain Adaption Network for Point Cloud Representation (NeurIPS 2019) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Domain Adaptation on Point Clouds via Geometry-Aware Implicits (CVPR 2022 · similarity 0.88)
- Deformation and Correspondence Aware Unsupervised Synthetic-to-Real Scene Flow Estimation for Point Clouds (CVPR 2022 · similarity 0.87)
- Meshing Point Clouds with Predicted Intrinsic-Extrinsic Ratio Guidance (ECCV 2020 · similarity 0.87)
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 2025 · similarity 0.87)
- Point Cloud Pre-training with Diffusion Models (CVPR 2024 · similarity 0.86)
- Sparse Point Cloud Patches Rendering via Splitting 2D Gaussians (CVPR 2025 · similarity 0.86)
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.
