Scribd
Upload
7 views

Deep Fake Detection

Uploaded by

Mariyam Amreen
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
7 views

Deep Fake Detection

Uploaded by

Mariyam Amreen
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 7

DEEP FAKE DETECTION

Presentation By : Mariyam Amreen


ID : 24007279004
Course: Research Methods

Submitted to : DR . DEC-2024
INTRODUCTIO
N
Deep Fakes are fake images and videos that are generated by
using deep learning techniques.

Deepfake detection refers to the identification and verification


of fake media such as images, videos, or audio that have been
altered using artificial intelligence techniques, particularly
deep learning. Deepfakes use algorithms like Generative
Adversarial Networks (GANs) to generate realistic but fake
content, making it difficult to differentiate between real and
fake media with the human eye.

The motivation for choosing this topic is from my personal


experience with online scams. I motivated to work on this
problem because deepfakes can harm everyday people very
easily.
LITERATURE TABLE
S.NO. Paper-Title Study Parameter Focus Finding Limitation Gap

01 SLR Detection methods High accuracy Lacks Need robust models


DL Methods Review (CNNs) generalization
02 Fairness in Detection Demographic bias Reducing bias Fairness with high Limited loss Broader testing
accuracy functions needed
03 Video Detection Challenges Model CNNs effective Dataset issues Real-time detection
Challenges generalization
04 Real-Time Video Eye movement Hybrid video 98.7% accuracy Poor live-streaming Live-stream
Detection detection adaptation
05 ELA + CNN Image analysis ELA preprocessing 89.5% accuracy Compressed media Generalize for video
issues
06 Biometrics Threats Biometric systems Risks of deepfakes Identity theft risks Weak benchmarks Real-world datasets

07 Benchmark Survey Generation/Detection Multimodal Improved models Dataset bias Real-time readiness
improvements
08 Detection Techniques CNNs outperform Poor scalability Diverse datasets
Review SLR Dataset evaluation ML needed
09 Generation/Detection Multimodal Integration of data High accuracy Dataset Real-time solutions
Study types generalization
10 Future Trends Trends Ethics and detection Blockchain Dataset gaps Scalable models
potential
LITERATURE MAP
PROBLEM STATEMENT

Deepfake technology threatens public trust, security, and


privacy, and despite achieving high accuracy, current
detection systems struggle with real-time recognition,
generalization, and evolving manipulation techniques,
necessitating advancements in multimodal detection,
improved dataset diversity, and scalable models to
prevent misuse in sensitive domains.
RESEARCH OBJECTIVES

• To develop a scalable and efficient deepfake detection


system by combining AI and blockchain that improves
real-time recognition.

• To enhance multimodal detection methods by


integrating diverse datasets and improving their
generalization across different deepfake scenarios.
THANK YOU

You might also like