ranahanocka/MeshCNN

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1731 stars · 342 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

MeshCNN is a PyTorch library implementing a convolutional neural network that operates directly on 3D triangular meshes, with mesh-specific convolution, pooling, and unpooling layers. It supports tasks like 3D shape classification and segmentation and was published at SIGGRAPH 2019.

Use cases

• classify 3D shapes with a neural network
• segment parts of a 3D mesh
• run deep learning on triangular meshes
• visualize learned mesh edge collapses
• train a CNN on SHREC or human segmentation datasets

When to choose

• you need deep learning directly on 3D mesh geometry rather than voxel or point-cloud representations
• you want a research-grade reference implementation of mesh convolutions in PyTorch
• you need pretrained models for 3D classification or segmentation benchmarks

When to avoid

• you work with point clouds or voxels instead of triangular meshes
• you need a maintained library with recent PyTorch support, since it targets PyTorch 1.2
• you need production-scale training pipelines rather than research code

Facets

library · maturity maintenance

deep-learning machine-learning image-processing deep-learning computer-vision graphics machine-learning python pytorch 3d-meshes geometric-deep-learning mesh-segmentation shape-classification siggraph linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ranahanocka/MeshCNN")