biometric-security

Bl-IEA is a research-based implementation of a bit-level fingerprint image encryption algorithm combining the Modified Pseudo Hadamard Transform (MPHT) and a Chaotic Skew Tent Map. It achieves strong diffusion, high randomness, and perfect reversibility, ensuring secure and efficient biometric data protection suitable for real-time applications.

Circular Hough transform is used for detecting circles in images. This is a basic implementation of CHT using Matlab

EfficientNet+MTCNN deepfake detection achieving 87% accuracy with 12.96% EER - Multi-modal video forensics for misinformation prevention

Banking-optimized signature verification with Siamese CNN - Contrastive learning for forgery detection with high recall for fraud prevention

Ensemble speaker verification achieving 97% accuracy - Intelligent fusion of MFCC+DTW (92%) and Resemblyzer CNN (94%) for voice authentication

This project implements an AI-powered Voice Word Lock system that authenticates users using a fixed spoken passphrase combined with speaker verification and voice biometrics. The system captures voice input, extracts biometric features, and verifies the speaker’s identity using machine learning models. Access is granted only when both the spoken

Jarvis AI Assistant is a comprehensive desktop application that transforms your computer into an intelligent, voice-controlled environment. Built with Python and modern web technologies, it provides hands-free access to system functions, health monitoring, information retrieval, and entertainment via natural voice commands and biometric security.

Multi-modal biometric access control combining DeepFace facial recognition and height verification for RBAC. Built on Johnson Controls Security Products internship work. ```

BiometricID — iOS SDK for face recognition using TrueDepth IR camera. On-device CoreML embedding generation ensures photos never leave the device. Dual-modal fusion (IR 70% + RGB 30%) delivers high accuracy. ECIES encryption and BioHash protect biometric data. Simple API: register and authenticate users in minutes.

Identify yourself with FaceID, using BiometricID Framework you can register any number of faces for feature login in your app. Don't use login and password, use your face for quick and safe login.

Face anti-spoofing system for biometric security using MobileNetV2, frequency cues, and rigorous video-level PAD evaluation.

Active liveness and deepfake detection pipeline using webcam red/blue light challenges, MediaPipe face landmarks, OpenCV, and temporal-spatial-frequency analysis for face anti-spoofing and presentation attack detection.

SecureVault is a face-authentication-based password manager that uses facial recognition to securely store and access credentials without traditional passwords. Built with Python, OpenCV, and Flask.

Free zero-trust broker for passwordless auth — Docker image, three SDKs, and docs.

Project Overview

Real-time presentation attack detection separating a live face from a photo, screen or mask.

Research framework for identity-invariant face anti-spoofing representation learning.

A real-time continuous authentication system using face recognition (FaceNet), keystroke dynamics (LSTM Autoencoder), and mouse dynamics (GRU Autoencoder) with a Chrome extension, FastAPI backend, and React admin portal.

FaceKYC hỗ trợ đối chiếu khuôn mặt 1:1 và sàng lọc presentation attack trong quy trình eKYC. Hệ thống giúp đội vận hành, rủi ro và tuân thủ ưu tiên hồ sơ cần chụp lại hoặc rà soát thủ công, đồng thời cung cấp API và dashboard để tích hợp, giám sát và quản trị mô hình.