Computer Vision & ML

Real-time Face Recognition & Liveness Detection

Real-time face detection and recognition pipeline using MTCNN and CNN fine-tuned on VGGFace2, featuring texture and motion-based liveness/anti-spoofing detection for secure biometric access.

Key Architectural & System Features

Built pipeline using MTCNN and transfer learning on VGGFace2 achieving 95.2% accuracy
Implemented liveness detection using texture and motion-based cues to prevent spoofing
Optimized CNN architecture for edge deployment reducing inference time by 67%

Technologies & Tools Stack

OpenCVTensorFlowMTCNNCNNYOLOVGGFace2