Attending Generalizability in Course of Deep Fake Detection by Exploring Multi-Task Learning

Pranav Balaji, Abhijit Das, Srijan Das, Antitza Dantcheva

ICCVW 2023 pp. 475-484

doi:10.1109/ICCVW60793.2023.00054 /iccvw/2023/balaji2023iccvw-attending/

Abstract

This work explores various ways of exploring multi-task learning (MTL) techniques aimed at classifying videos as original or manipulated in cross-manipulation scenario to attend generalizability in deep fake scenario. The dataset used in our evaluation is FaceForensics++, which features 1000 original videos manipulated by four different techniques, with a total of 5000 videos. We conduct extensive experiments on multi-task learning and contrastive techniques, which are well studied in literature for their generalization benefits. It can be concluded that the proposed detection model is quite generalized, i.e., accurately detects manipulation methods not encountered during training as compared to the state-of-the-art.

PDF ICCVW Semantic Scholar

Cite

Text

Balaji et al. "Attending Generalizability in Course of Deep Fake Detection by Exploring Multi-Task Learning." IEEE/CVF International Conference on Computer Vision Workshops, 2023. doi:10.1109/ICCVW60793.2023.00054

Markdown

[Balaji et al. "Attending Generalizability in Course of Deep Fake Detection by Exploring Multi-Task Learning." IEEE/CVF International Conference on Computer Vision Workshops, 2023.](https://mlanthology.org/iccvw/2023/balaji2023iccvw-attending/) doi:10.1109/ICCVW60793.2023.00054

BibTeX

@inproceedings{balaji2023iccvw-attending,
  title     = {{Attending Generalizability in Course of Deep Fake Detection by Exploring Multi-Task Learning}},
  author    = {Balaji, Pranav and Das, Abhijit and Das, Srijan and Dantcheva, Antitza},
  booktitle = {IEEE/CVF International Conference on Computer Vision Workshops},
  year      = {2023},
  pages     = {475-484},
  doi       = {10.1109/ICCVW60793.2023.00054},
  url       = {https://mlanthology.org/iccvw/2023/balaji2023iccvw-attending/}
}