Research map / Point clouds / Point cloud classification
SageMix: Saliency-Guided Mixup for Point Clouds
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 SageMix augmentation to training | PointNet++ · MN40, PB_T50_RS (PointNet++) | overall accuracy: 90.7 -> 93.3 | Helped |
| add SageMix augmentation to training | DGCNN · PB_T50_RS (DGCNN) | overall accuracy: 79.9 -> 83.6 | Helped |
| add SageMix augmentation to training | DGCNN · ECE on OBJ_ONLY | ECE (%): 19.8 -> 5.1 | Helped |
| add SageMix augmentation to training | DGCNN · OBJ_ONLY Gaussian noise 0.05 | accuracy: 48.4 -> 51.2 | Mixed |
| sample query points uniformly instead of none (Mixup baseline) (ablation) | SageMix · OBJ_ONLY | overall accuracy: 86.2 -> 86.8 | Helped |
| select query point deterministically by max saliency (ablation) | SageMix · OBJ_ONLY | overall accuracy: 86.2 -> 86.1 | Hurt |
| sample query points by saliency probability instead of uniform (ablation) | SageMix · OBJ_ONLY | overall accuracy: 86.8 -> 87.8 | Helped |
| add distance reweighting for second query point (ablation) | SageMix · OBJ_ONLY | overall accuracy: 87.8 -> 88.0 | Helped |
| vary RBF bandwidth sigma from 0.3 to 0.1 or 2.0 (ablation) | SageMix · OBJ_ONLY | overall accuracy: 88.0 -> 87.2 (0.1), 87.6 (2.0) | Hurt |
| apply SageMix to part segmentation | PointNet++ · SN-Parts | accuracy: 85.1 -> 85.7 | Helped |
| apply SageMix to 2D images versus Co-Mixup | PreActResNet18 · CIFAR-100 | accuracy: 80.13 -> 80.16 | No clear effect |
Closest papers by meaning
- PointPatchMix: Point Cloud Mixing with Patch Scoring (AAAI 2024 · similarity 0.90)
- Learning Continuous Implicit Field with Local Distance Indicator for Arbitrary-Scale Point Cloud Upsampling (AAAI 2024 · similarity 0.84)
- On Geometry-Enhanced Parameter-Efficient Fine-Tuning for 3D Scene Segmentation (NeurIPS 2025 · similarity 0.84)
- Point Cloud Pre-training with Diffusion Models (CVPR 2024 · similarity 0.83)
- Adversarial AutoMixup (ICLR 2024 · similarity 0.83)
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 2025 · similarity 0.83)
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.