
Private Virtual Stylist is a smart, on-device fashion assistant that allows users to virtually try on clothes and receive intelligent outfit recommendations—all while ensuring total privacy. Built with the HR-VITON pipeline for high-quality try-on synthesis, the system uses Grok to extract metadata from clothing images (e.g., type, color, pattern) and LLaMA to generate natural language fashion advice tailored to the user's current outfit or selected item. Running entirely offline on Qualcomm Snapdragon-powered Edge AI devices, this solution is optimized for privacy-first fashion experiences—ideal for mobile shopping, in-store kiosks, or use in regions with limited connectivity. Key Capabilities: • Virtual Try-On: Upload a photo and clothing item → see realistic try-on result • Style Intelligence: Metadata extraction via Grok + outfit advice from LLaMA • Offline-First: All inference runs on-device—no cloud, no data sharing • Privacy-First: Aligned with GDPR and secure by design Tech Stack Highlights: • HR-VITON for image-based try-on • Grok for clothing metadata parsing • LLaMA (LLM) for outfit suggestions and personalized recommendations • Pose estimation & image preprocessing for realism • Edge AI deployment using ONNX
8 Jul 2025