Pytorch source code for the paper:
M. Barni, R. Donida Labati, A. Genovese, V. Piuri, and F. Scotti,
"Iris deidentification with high visual realism for privacy protection on websites and social networks",
in IEEE Access, vol. 9, 2021, pp. 131995-132010. ISSN: 2169-3536.
[DOI: 10.1109/ACCESS.2021.3114588]
Article:
https://ieeexplore.ieee.org/document/9543669
Project page:
https://iebil.di.unimi.it/irisGan/irisGan.html
Outline:
Citation:
@Article {iride21,
author = {M. Barni and R. {Donida Labati} and A. Genovese and V. Piuri and F. Scotti},
title = {Iris deidentification with high visual realism for privacy protection on websites and social networks},
journal = {IEEE Access},
volume = {9},
pages = {131995-132010},
year = {2021},
note = {2169-3536},}
Main files:
- DCGAN-PyTorch_A_train: script that trains a GAN
- DCGAN-PyTorch_A_test: script that loads a trained GAN and generates synthetic textures
Required files:
- ./rsm/1/: Database of Rubber Sheet Models (RSM), with size 512x64, 8 bit (greyscale)
(some examples are already present)
Directories:
- ./images: directory containing the generated images
- ./models: directory containing the saved GAN models