AI Tremor-Print: Biometric Identification System (github.com)

A groundbreaking biometric identification system, named AI Tremor-Print, has been developed to leverage smartphone magnetometer sensors for capturing unique neuromuscular micro-tremors, facilitating individual identification. This innovative approach is particularly valuable for researchers, accessibility solutions, and cost-effective biometric authentication. By employing a smartphone's magnetometer, the system detects involuntary hand tremors, from which it extracts distinctive biometric signatures using optimized AI models, specifically a fine-tuned version of DistilGPT2. This enables real-time identity verification and offers a significant leap in low-cost biometric solutions, with processing entirely conducted on-device to ensure user privacy. The project showcases a fully integrated pipeline for data collection, AI training, and model verification, making it research-ready with comprehensive documentation for users. Key advantages include GPU optimization for consumer hardware, rapid identification times, and its applicability in monitoring neuromuscular disorders like Parkinson's disease. The method's reliance on naturally occurring tremors as biometric markers underscores its potential for not only accessibility technology but also for a variety of clinical applications. Overall, AI Tremor-Print represents a significant advancement in biometric systems, combining accessibility with potent AI-driven capabilities.