Import the HexMe dataset

[sebmestrallet]

As suggested in the review of our last article¹, we should evaluate our algorithm on the HexMe dataset²), which contains an extended set of inputs with interesting characteristics for hexahedral mesh generation. A comparison on HexMe would reveal how success rate/runtime/quality compare to state of the art frame-field based methods.

Here in dds-hexmeshing a new algorithm definition (like definitions/algorithms/import_HexMe.py) would fetch HexMe online, and insert it in the local data folder, under an appropriate file structure. Then definitions/algorithms/batch_processing.py would execute our labeling algorithm on these new CAD models.

Footnotes

1. S. Mestrallet, C. Bourcier, and F. Ledoux, "Validity-first automatic polycube labeling for CAD models", Proceedings of the 2025 International Meshing Roundtable, SIAM, March 2025, arXiv:2502.05712, GitHub:LIHPC-Computational-Geometry/validity-first-polycube-labeling ↩

2. P.-A. Beaufort, M. Reberol, D. Kalmykov, H. Liu, F. Ledoux, and D. Bommes, “Hex me if you can”, Computer Graphics Forum, vol. 41, no. 5, pp. 125–134, 2022, doi:10.1111/cgf.14608 ↩