A "minimal" package to compute self-intersections among various faces within a triangle-mesh.
This package currently provides a "CPU-only" version that can be used from PyTorch. Other enhancements, including a CUDA module, are under development.
Also, we currently do NOT implement a backward method, as this module is primarily intended for benchmarking convenience (i.e., to identify the self-intersections in a mesh generated by a neural network, for example).
Assuming you have a conda/virtual environment with PyTorch (>=1.3.0) installed, run the following command from the base directory of this repo (i.e., the directory containing this readme).
python setup.py build developA minimal example that uses the CPU version of the self-intersection test is shown below
# TODO: Add this