Getting started¶
did is a library which should be used for computing the DID dissimilarity between images.
This tutorial explains the installation of did as well as a first example.
Wait… why is the library called `did` and not `diffy` like the repo? We have not decided on a name yet… but you can open an issue if you have an opinion.
Installation¶
Dependencies¶
did has the following major dependencies. The versions given are those used to develop and test it, however they should be flexible. Please add an issue on Github if you notice dependency issues.
numpy and scipy
torch and torchvision
pillow
matplotlib
You can install these by hand with your choice of method (for example pip or conda), or use the environment.yml (conda create -n did_env -f environment.yml) and requirements.txt (pip install -r requirements.txt) available in the repository.
(did_env) $ pip install -r requirements.txt # method with pip
(base) $ conda create -n did_env -f environment.yml # method with conda
Locally installing the library (optional)¶
If you plan on using did in your own projects (and you should!), the easiest method is to locally install the package with python setup.py install in your environment.
This is not necessary if you simply want to reproduce the experiments in the paper (for instance, with the scripts in the repository).