Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Robust deepfake video detection using spatio-temporal features and dynamic difference learning.

Eman AbdElfattah¹, Hamdy M Mousa², Ashraf Elsisi²

¹Computer Science Department, Faculty of Computers and Information, Minufiya University, Shebin El Kom, Egypt. eman.abdelfattah@ci.menofia.edu.eg.

Scientific Reports

|June 2, 2026

Summary

Related Concept Videos

Difference from Background: Limit of Detection

Difference from Background: Limit of Detection

The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

A Secure and Efficient Framework for ECG Signal Digitization and Encryption for Internet of Medical Things Applications.

Healthcare technology letters·2026

Same author

A Photomicrographic Dataset of Rocks for the Accurate Classification of Minerals.

Scientific data·2025

Same author

Enhancing Heart Disease Diagnosis Using ECG Signal Reconstruction and Deep Transfer Learning Classification with Optional SVM Integration.

Diagnostics (Basel, Switzerland)·2025

Same author

Live Tracking and Dense Reconstruction for Handheld Monocular Endoscopy.

IEEE transactions on medical imaging·2018

Same author

On-patient see-through augmented reality based on visual SLAM.

International journal of computer assisted radiology and surgery·2016

This study introduces a deep learning framework for detecting deepfake videos by analyzing facial motion inconsistencies. The method achieves 100% accuracy on benchmark datasets, offering robust detection of manipulated facial content.

Area of Science:

Computer Science
Artificial Intelligence
Machine Learning

Background:

Deepfake technology advancements challenge traditional frame-by-frame video analysis.
Existing deepfake detection methods struggle with subtle spatial-temporal inconsistencies and facial motion irregularities.
Accurate detection of manipulated videos is crucial for combating misinformation.

Purpose of the Study:

To develop a comprehensive deep learning framework for enhanced deepfake video detection.
To address limitations in current methods by integrating spatial and temporal analysis with a focus on facial motion dynamics.
To improve the accuracy and robustness of deepfake detection systems.

Main Methods:

Facial landmarks extracted using Dlib's 68-point detector for geometric descriptors.

Keywords:

Deepfake detection Facial landmark detection Feature engineering Spatio-temporal analysis

Related Experiment Videos

Transformer encoder utilized to capture short- and long-term facial motion dynamics, enhanced by a Dynamic Difference Module (DDM).

Convolutional Neural Networks (CNNs) for spatial feature extraction and Long Short-Term Memory (LSTM) networks for temporal dependencies.

Main Results:

The proposed framework achieved 100% accuracy across three benchmark datasets: FaceForensics++ (FF++), UADFV, and DFDC.
Demonstrated superior performance in capturing subtle spatial-temporal inconsistencies and unnatural facial motion.
The Dynamic Difference Module effectively highlighted abrupt changes indicative of manipulation.

Conclusions:

The integrated deep learning framework provides a robust and highly accurate solution for deepfake video detection.
The approach effectively overcomes limitations of frame-level analysis by incorporating detailed facial motion dynamics.
The demonstrated 100% accuracy signifies strong generalization capabilities for real-world applications.