aidcam-tools

Small, dependency-light Python utilities for working with dental 3D files and evaluating AI dental CAD methods.

Released under MIT by the AID·CAM team.

---

Why this exists

Public benchmarks for AI dental CAD are slowly emerging — but standardized tooling to use them isn't. aidcam-tools is our small contribution: a set of plain Python scripts that we found ourselves writing for our own crown-generation work, packaged so anyone else working on dental ML can skip writing them too.

This repo is intentionally narrow. It contains no model code, no AID·CAM proprietary architecture, and no clinical data. It is dental file plumbing — the boring useful kind.

What's included

Module	What it does
aidcam_tools.mesh_to_pointcloud	Load STL/OBJ/PLY meshes via trimesh; uniformly sample N points from the surface; save to .npz.
aidcam_tools.metrics	Chamfer distance (L1, L2), symmetric Hausdorff distance, per-tooth-class breakdown for FDI numbering.
aidcam_tools.fdi	FDI tooth-numbering helpers: quadrant, position, is_molar, is_anterior, opposite, contralateral.
aidcam_tools.eval_intellident	Evaluation harness for the public 78-case margin-line benchmark from Intellident Dentaire (CBM 2025).

Install

pip install -e .

Requires Python ≥ 3.10. Dependencies: numpy, trimesh, scipy.

Examples

Convert a dental mesh to a point cloud

aidcam-mesh-to-pc input_prep.stl prep_points.npz --n-points 10240

Compare two crown meshes

import numpy as np
from aidcam_tools.metrics import chamfer_l2, hausdorff

a = np.load("crown_a.npz")["points"]
b = np.load("crown_b.npz")["points"]

print(f"Chamfer L2: {chamfer_l2(a, b):.4f} mm")
print(f"Hausdorff: {hausdorff(a, b):.4f} mm")

Work with FDI tooth numbers

from aidcam_tools.fdi import is_molar, opposite, quadrant

is_molar(16)       # True
opposite(11)       # 41  (upper right central → lower right central)
quadrant(35)       # 3   (lower-left quadrant)

Evaluate predicted margin lines against the Intellident benchmark

git clone https://github.com/intellident-ai/public-datasets
aidcam-eval-intellident \
  --benchmark public-datasets/ \
  --predictions my_predictions/ \
  --output report.json

License

MIT. See LICENSE.

Citation

If you use these utilities in published work, a citation is appreciated but not required:

@software{aidcam_tools_2026,
  author       = {AID·CAM},
  title        = {aidcam-tools: utilities for AI dental CAD},
  year         = {2026},
  url          = {https://github.com/AID-CAM/aidcam-tools},
}

Contributing

Issues and pull requests welcome. The scope of this repo is intentionally narrow — utility scripts for dental ML. We're not looking to grow it into a full framework. If your contribution fits the "small dependency- light tools" bar, please open an issue first to discuss.

Maintainers

• @crosswi — Cameron Cross, AID·CAM
• @massimovenezia — Massimo Venezia, AID·CAM

For questions, contact info@aid-cam.com.