Research map / Point clouds / Point cloud classification
MeshMAE: Masked Autoencoders for 3D Mesh Data Analysis
Yaqian Liang, Shanshan Zhao, Baosheng Yu, Jing Zhang, Fazhi He
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 MAE pre-training on ShapeNet | Mesh Transformer · Manifold40 classification | accuracy: 91.5 -> 92.5 | Improved |
| add MAE pre-training on ModelNet40 | Mesh Transformer · Manifold40 classification | accuracy: 91.5 -> 91.7 | Improved |
| pre-train on larger ShapeNet instead of ModelNet40 | MeshMAE · Manifold40 classification | accuracy: 91.7 -> 92.5 | Improved |
| add pre-training | Mesh Transformer · COSEG-aliens segmentation | accuracy: 97.6 -> 98.0 | Improved |
| add pre-training on ModelNet40 | Mesh Transformer · Human Body segmentation | accuracy: 90.1 -> 90.1 | No clear effect |
| add the Chamfer vertex loss to the MSE face-feature loss (weight 0.5) (ablation) | MeshMAE · ModelNet40 fine-tuning | accuracy: 91.2 -> 91.7 | Improved |
| add the Chamfer vertex loss (weight 0.5) in linear probing (ablation) | MeshMAE · ModelNet40 linear probing | accuracy: 89.8 -> 91.1 | Improved |
| add MSE face-feature loss to the Chamfer loss alone (ablation) | MeshMAE · ModelNet40 fine-tuning | accuracy: 90.6 -> 91.7 | Improved |
| raise Chamfer loss weight from 0.5 to 2 (ablation) | MeshMAE · ModelNet40 linear probing | accuracy: 91.1 -> 89.7 | Hurt |
| randomize the order of face features in a patch (ablation) | MeshMAE · ModelNet40 fine-tuning | accuracy: 91.7 -> 91.6 | No clear effect |
| embed the patch center instead of using learnable positional parameters (ablation) | MeshMAE · ModelNet40 linear probing | accuracy: 46.1 -> 89.7 | Improved |
| embed the patch center instead of reshaping 64 face centers (ablation) | MeshMAE · ModelNet40 linear probing | accuracy: 77.5 -> 89.7 | Improved |
Closest papers by meaning
- MeshFormer : High-Quality Mesh Generation with 3D-Guided Reconstruction Model (NeurIPS 2024 · similarity 0.87)
- Towards Compact 3D Representations via Point Feature Enhancement Masked Autoencoders (AAAI 2024 · similarity 0.87)
- MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · similarity 0.86)
- PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers (NeurIPS 2025 · similarity 0.86)
- ARMesh: Autoregressive Mesh Generation via Next-Level-of-Detail Prediction (NeurIPS 2025 · similarity 0.86)
- 3D Feature Prediction for Masked-AutoEncoder-Based Point Cloud Pretraining (ICLR 2024 · 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.
