TRUEVISION: Vision Based Deepfake Detection System

DOI

10.2991/978-94-6463-831-8_46How to use a DOI?

Keywords

Deep Fake Detection; Vision Based Analysis; Real-time Detection; Continuous Learning; BI; GAN; Grad-CAM; MCL; SOTA; NFA; Cosine Annealing; AdamW; Mediapipe

Abstract

As deep fake generation techniques evolve, detecting manipulated content across multiple modalities has become increasingly important to safeguard against their misuse. This paper presents a vision based deep fake detection system that leverages the power of pre-trained transformer models to analyze two different data modalities, facial images. For facial images, an image transformer captures pixel-level, structural and spatial inconsistencies. We discuss the integration of these transformers within a unified system, the synergistic effect of vision-based analysis, and the improvement in performance compared to CNN based approaches providing insights into potential directions for further enhancing the detection of sophisticated deep fakes.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Amey Surendra Datar
AU  - Atharv Abhijit Warkari
AU  - Prathmesh Prafulla Patil
AU  - Jagruti A. Wagh
PY  - 2025
DA  - 2025/08/31
TI  - TRUEVISION: Vision Based Deepfake Detection System
BT  - Proceeding of the 1st International Conference on Lifespan Innovation (ICLI 2025)
PB  - Atlantis Press
SP  - 379
EP  - 387
SN  - 2468-5739
UR  - https://doi.org/10.2991/978-94-6463-831-8_46
DO  - 10.2991/978-94-6463-831-8_46
ID  - Datar2025
ER  -