Research map / Point clouds / Point cloud classification
Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data
Mikaela Angelina Uy, Quang-Hieu Pham, Binh-Son Hua, Thanh Nguyen, Sai-Kit Yeung
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 a segmentation-guided background-aware branch (ablation) | PointNet++ · ScanObjectNN PB_T50_RS | overall accuracy (%): 77.9 -> 80.2 | Improved |
| add a segmentation-guided background-aware branch (ablation) | DGCNN · ScanObjectNN PB_T50_RS | overall accuracy (%): 78.1 -> 79.7 | Improved |
| add a segmentation-guided background-aware branch (ablation) | PointNet++ · ModelNet40 (trained on PB_T50_RS) | overall accuracy (%): 47.4 -> 52.6 | Improved |
| add a segmentation-guided background-aware branch (ablation) | DGCNN · ModelNet40 (trained on PB_T50_RS) | overall accuracy (%): 54.7 -> 56.5 | Improved |
| train on real-world ScanObjectNN OBJ_ONLY instead of ModelNet40 | PointNet++ · ScanObjectNN OBJ_ONLY | overall accuracy (%): 43.6 -> 84.3 | Improved |
| add background points to objects (OBJ_BG vs OBJ_ONLY) | PointNet++ · ScanObjectNN | overall accuracy (%): 84.3 -> 82.3 | Hurt |
| add background points to objects (OBJ_BG vs OBJ_ONLY) | PointCNN · ScanObjectNN | overall accuracy (%): 85.5 -> 86.1 | Improved |
| add background points in hardest variant PB_T50_RS | DGCNN · ScanObjectNN PB_T50_RS | overall accuracy (%): 81.5 -> 78.1 | Hurt |
| add background points in training for ModelNet40 test | PointNet · ModelNet40 (trained on PB_T50_RS) | overall accuracy (%): 60.4 -> 50.9 | Hurt |
| add background points in training for ModelNet40 test | PointCNN · ModelNet40 (trained on PB_T50_RS) | overall accuracy (%): 38.1 -> 49.2 | Improved |
| add translation, rotation and scale perturbations (PB_T50_RS vs OBJ_ONLY) | DGCNN · ScanObjectNN | overall accuracy (%): 86.2 -> 78.1 | Hurt |
| train on ScanObjectNN instead of ModelNet40 for testing on ModelNet40 | PointNet++ · ModelNet40 | overall accuracy (%): 47.4 -> 55.0 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PointCNN — PointCNN: Convolution On X-Transformed Points (NeurIPS 2018) · better on 2 of 2 shared results
- SpiderCNN — SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filters (ECCV 2018) · 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
- Deep Learning for 3D Point Clouds: A Survey (TPAMI 2020 · similarity 0.90)
- Revisiting Point Cloud Shape Classification with a Simple and Effective Baseline (ICML 2021 · similarity 0.90)
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (CVPR 2017 · similarity 0.89)
- LiDAR-Net: A Real-scanned 3D Point Cloud Dataset for Indoor Scenes (CVPR 2024 · similarity 0.89)
- Deep Hough Voting for 3D Object Detection in Point Clouds (ICCV 2019 · similarity 0.88)
- DeepLA-Net: Very Deep Local Aggregation Networks for Point Cloud Analysis (CVPR 2025 · 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.
