Research map / Point clouds / Point cloud classification
PointMixer: MLP-Mixer for Point Cloud Understanding
Jaesung Choe, Chunghyun Park, Francois Rameau, Jaesik Park, In So Kweon
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 relative positional encoding (ablation) | PointMixer · S3DIS Area 5 | mIoU: 64.7 -> 71.4 | Improved |
| add hierarchical-set mixing (symmetric up/down layers) (ablation) | PointMixer · S3DIS Area 5 | mIoU: 69.1 -> 71.4 | Improved |
| add hierarchical-set mixing (symmetric up/down layers) (ablation) | PointMixer · ShapeNet-Part reconstruction | Chamfer distance: 1.12 -> 1.11 | Improved |
| combine hierarchical-set and inter-set mixing (ablation) | PointMixer · S3DIS Area 5 | mIoU: +2.2 mIoU | Improved |
| combine hierarchical-set and inter-set mixing (ablation) | PointMixer · ShapeNet-Part reconstruction | Chamfer distance: +0.02 | Improved |
| add softmax weighting to channel MLPs (ablation) | PointMixer · S3DIS Area 5 | mIoU: 58.3 -> 71.4 | Improved |
| add token-mixing MLPs on top of softmax (ablation) | PointMixer · S3DIS Area 5 | mIoU: 71.4 -> 71.1 | No clear effect |
| use PointMixer instead of Point Transformer | PointMixer · S3DIS Area 5 | mIoU: 70.4 -> 71.4 | Improved |
| use PointMixer instead of Point Transformer | PointMixer · ScanNet reconstruction | Chamfer distance: 2.79 -> 2.74 | Improved |
| use PointMixer instead of Point Transformer | PointMixer · ICL-NUIM reconstruction | F1: 41.6 -> 44.7 | Improved |
| use PointMixer instead of Point Transformer | PointMixer · ModelNet40 | OA: 93.7 -> 93.6 | No clear effect |
| use PointMixer instead of Point Transformer | PointMixer · ModelNet40 | mAcc: 90.6 -> 91.4 | Improved |
Closest papers by meaning
- PointPatchMix: Point Cloud Mixing with Patch Scoring (AAAI 2024 · similarity 0.89)
- Point MixSwap: Attentional Point Cloud Mixing via Swapping Matched Structural Divisions (ECCV 2022 · similarity 0.88)
- PointMixup: Augmentation for Point Clouds (ECCV 2020 · similarity 0.88)
- Improved MLP Point Cloud Processing with High-Dimensional Positional Encoding (AAAI 2024 · similarity 0.86)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.86)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 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.
