🛡️ DeepFake Detector

A forensic-grade deepfake detection toolkit designed for law enforcement, military, and national security applications. Built with Ruby and featuring a modern web interface, comprehensive CLI, and court-admissible reporting capabilities.

🇹🇷 Created by Ahmet KAHRAMAN (@ahmetxhero)
🔒 "Security first, innovation always!"

The DeepFake Detector provides reliable, forensic-grade analysis of manipulated media (deepfakes) with court-admissible evidence generation. It supports both real-time detection for field operations and comprehensive forensic analysis for investigations.

• Video: MP4, AVI, MOV, MKV
• Audio: WAV, MP3, FLAC
• Images: PNG, JPEG, BMP, RAW
• Live Streams: RTSP/HTTP feeds for real-time monitoring

• Visual Analysis: Facial landmark analysis, lighting consistency, pixel-level anomaly detection
• Audio Analysis: Spectrogram anomaly detection, voice synthesis detection
• Cross-Modal Verification: Lip-sync analysis, temporal consistency checks

• Authenticity Scores: Numeric confidence (0.00 – 1.00)
• Multiple Formats: JSON (machine-readable), PDF (court-ready), HTML (field use)
• Chain of Custody: SHA-256/SHA-3 hashing, cryptographic signing
• Audit Trail: Immutable logs of all operations

• Role-based access control (RBAC)
• GDPR/KVKK compliant data handling
• Offline capability for air-gapped systems
• Evidence integrity verification

• Accuracy: ≥ 95% detection accuracy on benchmark datasets
• Latency: ≤ 500ms/frame for real-time detection
• Scalability: Terabyte-scale dataset processing
• Deployment: Docker containers, Kubernetes clusters

# Install dependencies
bundle install

# Setup database
rake db:setup

# Run tests
rspec

# Start the service
bundle exec puma -C config/puma.rb

# Analyze a single file
./bin/deepfake-detector analyze /path/to/media.mp4

# Batch processing
./bin/deepfake-detector batch /path/to/directory

# Real-time stream analysis
./bin/deepfake-detector stream rtsp://camera.url

# Health check
curl http://localhost:9292/health

# Upload and analyze media
curl -X POST -F "file=@media.mp4" http://localhost:9292/api/v1/analyze

# Get analysis results
curl http://localhost:9292/api/v1/results/{analysis_id}

• Core Engine: Ruby-based CLI + REST API service
• AI Models: ONNX format for GPU/CPU compatibility
• Database: PostgreSQL for production, SQLite for development
• Security: Cryptographic signing, access control, audit logging

• v0.1: Prototype with basic video/image detection
• v0.2: Audio analysis and cross-modal verification
• v0.3: Chain of custody and signed reports
• v0.4: Real-time streaming and dashboard
• v1.0: Law enforcement ready release

• Probabilistic output requires human analyst verification
• GDPR/KVKK compliant personal data processing
• Comprehensive audit logging to prevent misuse
• Documented training datasets to mitigate bias

Proprietary - For authorized law enforcement and security agencies only.