Abstract :

Synthetic Media Enabled By Generative AI Has Become Increasingly Realistic Making It Highly Challenging To Differentiate Altered Images, Video, And Audio. The Majority Of The Existing Detection Algorithms Only Classify Alterations In A Single Modality At A Time, Limiting The System To Identify Partially Altered Content, Such As A Real Image With Fake Audio Or A Fake Video With Real Audio. The Paper Proposes "DeepGuard" A Comprehensive Multimedia Fraud Detection System For Images, Videos And Audio. The System Consists Of A Cross-modal Consistency Module To Evaluate The Consistency Of Visual And Audio Segments Of A Video. The System Uses EfficientNetB0 To Analyze Both Images And Videos, And It Uses A Four-layer CNN On Mel Spectrograms To Classify Audio. Analyses Of The Video Results Are Integrated With The Audio Analysis For Each Video And Are Processed Using A Lip-synchronization Test To Assess The Category Of Manipulation That Occurred. Experimental Results On Benchmark Datasets Achieved Validation Accuracies Of 98.17% For Image Detection, 98.40% For Video Detection, And 99.86% For Audio Detection. The System Also Generates A Timestamped Forensic Report Summarizing The Forgeries Detected And Suggested Actions.

Published:

15-5-2026

Issue:

Vol. 26 No. 5 (2026)

Page Nos:

1212-1219

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite