DeepFake Detection System

The rise of DeepFake technology poses a significant threat to the authenticity of visual media content. This project presents a novel approach for detecting DeepFake videos using a Convolutional Neural Network (CNN) enhanced with Long Short-Term Memory (LSTM) and ResNet50 architectures.

Leveraging the spatial and temporal features captured by CNNs, LSTM, and ResNet50, our model achieves high accuracy of 90.16% in discerning manipulated content. Through extensive experimentation on diverse datasets, we demonstrate the effectiveness of our approach in identifying subtle anomalies introduced by DeepFake generation methods.

The proposed detection system offers a scalable and efficient solution for real-time identification of DeepFake content in an era where visual media authenticity is of paramount importance.

CNN-LSTM Hybrid Architecture

Our hybrid architecture combines the power of Convolutional Neural Networks for spatial feature extraction with Long Short-Term Memory networks for temporal sequence analysis, creating a robust detection system.

Our innovative model integrates both CNN and RNN architectures for superior performance:

• 🧠
  ResNext CNN: Pre-trained model for robust frame-level feature extraction
• 🔗
  LSTM Network: Processes sequential features for temporal analysis
• ⚡
  Data Loader: Efficient training data pipeline with label management
• 🎯
  Binary Classification: Distinguishes between authentic and deepfake content