Skip to main navigation Skip to search Skip to main content

A Lightweight Framework for Emotion Drift and Asymmetry Detection in Explainable Deepfake Analysis

  • Shruthi Vishwajeeth*
  • , Balachandra
  • , Dayakshini
  • , Santhosha Rao
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Deepfakes are AI-generated synthetic media created through Generative Adversarial Networks (GANs) that mimic real human expressions with high realism, posing major security and ethical threats. Traditional detectors, primarily CNN and landmark-based, struggle with subtle manipulations and lack interpretability. This paper proposes a lightweight and explainable framework that fuses emotion drift analysis, facial asymmetry detection, and heatmap-based interpretability. By analyzing both behavioral (temporal emotion drift) and structural (asymmetry) cues, this framework improves explainability and robustness.

Original languageEnglish
Title of host publication2026 International Conference on Artificial Intelligence and Data Engineering, AIDE 2026 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages234-238
Number of pages5
ISBN (Electronic)9798331592288
DOIs
Publication statusPublished - 2026
Event2026 International Conference on Artificial Intelligence and Data Engineering, AIDE 2026 - Nitte, India
Duration: 05-02-202607-02-2026

Publication series

Name2026 International Conference on Artificial Intelligence and Data Engineering, AIDE 2026 - Proceedings

Conference

Conference2026 International Conference on Artificial Intelligence and Data Engineering, AIDE 2026
Country/TerritoryIndia
CityNitte
Period05-02-2607-02-26

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Computer Networks and Communications
  • Computer Science Applications
  • Computer Vision and Pattern Recognition
  • Information Systems
  • Statistics, Probability and Uncertainty

Fingerprint

Dive into the research topics of 'A Lightweight Framework for Emotion Drift and Asymmetry Detection in Explainable Deepfake Analysis'. Together they form a unique fingerprint.

Cite this