FEERCI is an opinionated, easy-to-use package for calculating EERs and non-parametric confidence intervals efficiently. This repository contains scripts for running speed and accuracy benchmarks against a similar function in the bob toolkit. We provide a docker container which this benchmark can be run in.
The easiest way to run the benchmark is using Docker:
docker build -t feerci_benchmark . docker run feerci_benchmark > results.csv
Otherwise, please ensure you are running an environment with bob installed. After this, run:
pip install -r requirements.txt ./benchmark_run.sh > results.csv
We provide a script for plotting, it assumes the existence of results.csv in the working directory:
pip install -r requirements_plots.txt python benchmark_plots.py