
Deepfake Face Detection System
A specialized platform engineered to detect deep fake pictures, morphed images, and synthetic or AI-manipulated video frames. The system applies a binary classification model to distinguish between authentic and manipulated facial media. For video streams, it sequentially inspects frames, aggregates confidence scores, and highlights segments that appear suspicious.
Distinguishes authentic from manipulated faces with confidence scoring.
Sequentially parses MP4 frames to isolate high-risk video segments.
Visualises spatial facial regions influencing model predictions.
Resilience testing across JPEG compression and Gaussian blur.
