Back to Projects
AI-Driven Facial Recognition & Anti-Spoofing Ecosystem
PythonFlaskPyTorchInsightFaceONNX RuntimeOpenVINOOpenCVMySQL
Overview
A high-performance biometric recognition and attendance system designed for seamless enterprise integration. The core engine is built on a sophisticated Python stack, leveraging InsightFace for high-precision facial analysis and PyTorch for advanced anti-spoofing detection to prevent fraudulent entries via mobile screens or printed photos. Architected with a Flask-based API layer, the system provides real-time, high-FPS inference and has been successfully validated through rigorous field testing. A key technical milestone was the successful integration into a broader Smart Parking Management system, enabling simultaneous driver verification and automated attendance tracking synchronized directly with corporate ERP databases.
Key Features
Advanced Anti-Spoofing Protection using PyTorch neural networks
High-Performance Real-time Recognition with ONNX Runtime and OpenVINO
Enterprise ERP Integration via Flask-SQLAlchemy for attendance sync
High-Speed Biometric Enrollment workflow
Scalable Microservice Architecture for easy ecosystem integration
Real-time RTSP camera stream processing
Automated attendance tracking synchronized with ERP
Live vs spoof detection (screens/photos)
Achievements
Successfully distinguished live subjects from high-resolution spoofs
Achieved stable FPS and ultra-low latency for instant biometric verification
Seamlessly integrated as a standalone service into the SOF Parking suite
Validated through rigorous field testing by enterprise clients
Tech Stack
PythonFlaskPyTorchInsightFaceONNX RuntimeOpenVINOOpenCVMySQL