The code was tested with docker environment based on nvidia container for pytorch release 18.09.
with the following installations:
- python 3.6.5
- pytorch 0.5.0a0 (higher version such as 1.0.0 or 1.4.0 are expected to work as well)
- Numpy 1.19.4
- scikit-image 0.15.0
- OpenCV 4.1.0 (we are only using basic functions, other versions may just as well)
- Scipy 1.1.0
- Pillow 6.1.0
- tqdm 4.25.0
and for the generation of K function maps for training: - R 4.0.3 (with the spatstat package installed)
- Pandas 1.1.5
- pyper
- rpy2
The docker environment is available here:
docker pull shahira/pytorch_plus_r