Lightweight CUDA tensor functions with a focus on image data augmentation.
augpy is a lightweight library with minimal dependencies that provides CUDA tensor functions with a focus on image data augmentation. It uses the dlpack standard to export tensors to other libraries, such as Pytorch, with zero copying.
Make sure you have the following installed.
Compiler with C++14 support (e.g. GCC 5)
Cuda 10 or higher
Cmake 3.13 or higher
Now simply run setup.py as normal to build the wheel and install with pip.
python setup.py bdist_wheel
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.