FitSpaceAI: A Biometric-Secured Real-Time Virtual Try- On System for AI-Driven Design Preview
Year: 2026
Type: article
Abstract: FitSpaceAI is a pioneering virtual try-on application that leverages real-time AI-based face detection to deliver a seamless augmented reality (AR) shopping experience. Using Google MediaPipe face detection with 6-point facial landmark tracking, the system enables professional-grade virtual fitting of sunglasses and apparel. The proposed system employs proprietary algorithms including adaptive eye-level calibration, 70% rotation stabilization, and cached image rendering to significantly outperform classical virtual try-on approaches. The system operates at 30–60 FPS with sub-100 ms latency using a React.js frontend, HTML5 Canvas API, and WebRTC. Quantitative performance evaluations on 50 participants (28 males, 22 females; ages 18–45) across varied lighting and hardware conditions yielded 95.2% eye position tracking accuracy (±1.4%), 92.8% rotation tracking (±2.1%), and 89.7% real- time positional stability (±2.8%)—surpassing current industry standards. Mean user satisfaction: 4.6/5.0 (N=50, σ =0.31), SUS score 82.4 (Excellent). Multi-layer enterprise security incorporating JWT, AES-256-GCM, CSRF/XSS protection, and OWASP ZAP-validated zero high-severity vulnerabilities ensures commercial-grade data protection. Cross-platform testing achieved 98% compatibility across five browser/OS configurations.
Institution: Thakur College of Engineering and Technology
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7136129716
FWCI:
Cites:
Cited by:
Related to: