Source ID Tracker: Source Face Identity Protection in Face Swapping in IEEE International Conference on Multimedia and Expo (ICME), 2022.
Extracting face images with MTCNN:
python ext_ff.py
Generating face landmarks with dlib:
python save_dlib_mask/generate_landmarks_dlib_ff_df_save_img.py
Splitting the dataset and getting the txt:
python get_txt.py
Training the Source-ID-Tracker model:
python train.py
- Revealing source images from fake images
python generate_test_images.py
python calculate_id_sim.py
python calculate_metric.py
python calculate_id_sim_perturbation.py
- Revealing source images from fake images of simulated distortions
python generate_test_images.py
python generate_perturbation.py
python Revealed_perturbation.py
python calculate_id_sim_perturbation.py
@inproceedings{lin2022source,
title = {Source-{{ID-Tracker}}: {{Source Face Identity Protection}} in {{Face Swapping}}},
shorttitle = {Source-{{ID-Tracker}}},
booktitle = {2022 {{IEEE International Conference}} on {{Multimedia}} and {{Expo}} ({{ICME}})},
author = {Lin, Yuzhen and Chen, Han and Maiorana, Emanuele and Campisi, Patrizio and Li, Bin},
year = {2022},
pages = {01--06},
issn = {1945-788X}
}