Unofficial pytorch reimplementation of

SER-FIQ: Unsupervised Estimation of Face Image Quality Based on Stochastic Embedding Robustness

CVPR 2020

Authors:
Philipp Terhörst, Jan Niklas Kolf, Naser Damer, Florian Kirchbuchner, Arjan Kuijper

Reimplementation

SER-FIQ is implemented for iResNet architecture, pre-trained iResNet18 and iResNet50 are provided. The scaling of the SER-FIQ scores is different from the original implementation and needs to be adapted to the dataset/model.

Run

To run the example code,

1. download the model checkpoints
2. create a folder checkpoints and copy the file into it
3. Execure the example code with

   python main.py
   

Dependencies

Tested with pytorch 2.0, torchvision.

License

This project is licensed under the terms of the Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.