Deepfake Detection Using Graph Convolutional Networks (GCN)

Divya Samad, Kailash Chandra Bandhu

2025

Abstract

In recent years, the rise of deepfake content-digitally altered media that convincingly manipulates appearances or voices-has posed significant challenges across social, ethical, and cybersecurity landscapes. Traditional deepfake detection methods, primarily relying on Convolutional Neural Networks (CNNs), often struggle with capturing subtle facial irregularities and spatial relationships in fine detail. To address this, we propose a hybrid model that combines Graph Convolutional Networks (GCNs) and CNNs. Our model leverages GCNs to analyze facial landmarks as graphs, capturing relational information, while CNNs focus on pixel-level details within images. By merging outputs from both models, we create a robust approach that capitalizes on the strengths of each. We evaluate our method on a comprehensive deepfake dataset, showing improved accuracy over traditional CNN-based approaches, particularly in identifying nuanced manipulations. This research contributes a unique hybrid framework to enhance the reliability of deepfake detection.

Paper Citation

in Harvard Style

Samad D. and Chandra Bandhu K. (2025). Deepfake Detection Using Graph Convolutional Networks (GCN). In Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH; ISBN 978-989-758-784-9, SciTePress, pages 9-16. DOI: 10.5220/0014265200004928

in Bibtex Style

@conference{ritech25,
author={Divya Samad and Kailash Chandra Bandhu},
title={Deepfake Detection Using Graph Convolutional Networks (GCN)},
booktitle={Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH},
year={2025},
pages={9-16},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0014265200004928},
isbn={978-989-758-784-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH
TI - Deepfake Detection Using Graph Convolutional Networks (GCN)
SN - 978-989-758-784-9
AU - Samad D.
AU - Chandra Bandhu K.
PY - 2025
SP - 9
EP - 16
DO - 10.5220/0014265200004928
PB - SciTePress