FaceNet with memory vs. synthetic faces
This repo contains code and data needed to re-produce the results presented in this blog post.
The repo does not, however, contain the weights of the FaceNet model. You can download them using the links provided in the utils.py file.
The memory.py file was copied from
this repo, and is an implementation of a
memory module by Kaiser et al.
If you have conda installed, you can create the environment by calling:
conda env create -f environment.yml
which will build for you the facememory conda environment which you can activate with:
source activate facememory
or
conda activate facememory
depending on which version of conda you're using.
But before you'll be ready to run the results.ipynb notebook, you also need to prepare the facenet submodule:
git submodule update
And now you're ready to run the notebook and get the results!
If you encounter any problems with the environment, or find a bug in the code, please let us know via an Issue.