Python code to test NIST frvt accuracy of MUGSHOT & pnas datasets

cd 11
./compileAndRunFrvt.sh
pip install -r requirements.txt
python plot.py

• NIST comes with the MUGSHOT dataset for testing:
  • 653 pairs (1306 img ppm files)
• pnas is also used for quick verification:
  • 20 pairs (12 Genuine(G) pairs & 8 Imposter(I) pairs)
• Additional tensorflow FR model & lib/*.so need to be downloaded seperately:
  • 09-02_02-45.pb

G-I Similarity box scatter chart

• MUGSHOT:

• pnas:

Face Recognition Vendor Test (FRVT) Validation Packages

This repository contains validation packages for all Ongoing FRVT evaluation tracks. We recommend developers clone the entire repository and run validation from within the folder that corresponds to the evaluation of interest. The ./common directory contains files that are shared across all validation packages.