130K Real vs Fake Face Dataset

Dataset Overview

Dataset Breakdown

1. Real Face Images (FFHQ Subset)

• Quantity: 70,000 images
• Resolution: 256×256 pixels
• Features:
  • High-quality real-world facial images
  • Diverse in age, ethnicity, gender, and pose
  • Suitable for facial recognition, age estimation, and AI bias detection

2. Synthetic Face Images (AI-Generated)

• Quantity: 60,000 high-resolution synthetic images
• Resolution: 1024×1024 pixels
• Features:
  • Created using Flux1.pro, Flux1.dev, and SDXL (Stable Diffusion XL)
  • High-resolution, realistic AI-generated facial images
  • Perfect for deepfake detection and AI-generated content research

Key Advantages of This Dataset

• Perfect for AI & ML Training – Train deep learning models to distinguish real vs fake faces.
• Supports Deepfake Detection Research – Enhance security against AI-generated face fraud.
• High-Resolution AI Faces – 1024×1024 synthetic images improve training accuracy.
• Diverse & Balanced Data – Includes real and fake faces for unbiased AI development.
• Ideal for Computer Vision & AI Ethics – Study synthetic media and its impact on real-world applications.

Applications of the 130K Real vs Fake Face Dataset

• Deepfake Detection: Build AI models to detect manipulated images.
• Facial Recognition: Improve identity verification and security systems.
• AI Ethics Research: Study biases in synthetic media generation.
• Computer Vision Projects: Enhance image classification and facial authentication.
• Adversarial Machine Learning: Test AI robustness against fake images.

Why Choose This Dataset?

This dataset is ideal for researchers, AI developers, and security experts working on face authentication, deepfake prevention, and AI-generated media detection.