Related Experiment Video
Updated: Sep 14, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
637
Design and development of an efficient RLNet prediction model for deepfake video detection
Varad Bhandarkawthekar1, T M Navamani1, Rishabh Sharma1
1School of Computer Science and Engineering (SCOPE), Vellore Institute of Technology (VIT), Vellore, Tamil Nadu, India.
Frontiers in Big Data
|July 24, 2025
Summary
This study introduces RLNet, a deep learning framework combining ResNet and Long Short Term Memory (LSTM) networks for accurate deepfake video detection. The model achieves 95.2% accuracy by analyzing both spatial and temporal features, outperforming existing methods.
Area of Science:
- Computer Science
- Artificial Intelligence
- Digital Forensics
Background:
- Deepfake videos pose significant security risks due to sophisticated forgery techniques.
- Existing detection methods often focus on spatial features, neglecting crucial temporal information.
- Developing robust deepfake detection is essential for digital content authenticity.
Purpose of the Study:
- To introduce RLNet, a deep learning framework for high-precision deepfake video detection.
- To leverage both spatial and temporal features for accurate identification of manipulated content.
- To enhance the security and integrity of digital media.
Main Methods:
- Utilized a deep learning framework (RLNet) integrating ResNet and Long Short Term Memory (LSTM) networks.
- Preprocessed diverse datasets containing authentic and deepfake videos.
- ResNet captured frame-level spatial anomalies; LSTM analyzed temporal inconsistencies across sequences.
Main Results:
- Achieved 95.2% accuracy in deepfake video detection.
- Demonstrated superior performance compared to existing methods like EfficientNet and Recurrent Neural Networks (RNN).
- The framework proved versatile and robust across various deepfake techniques and compression levels.
Conclusions:
- The combined ResNet and LSTM approach effectively detects deepfake videos by analyzing spatial and temporal features.
- RLNet offers a robust and versatile tool for digital media forensics.
- This research significantly contributes to enhancing the security and integrity of digital content.
