Inspiration

Daily routines are fragmented across multiple devices—phones for reminders, apps for weather, mirrors for appearance. The idea was to merge these into a single, intuitive interface that enhances everyday life without adding friction. The goal: turn a passive object (mirror) into an intelligent assistant.

What it does

Smart Mirror is an interactive AI-powered system that combines personal assistance with visual feedback. It:

•⁠ ⁠Suggests outfits using a virtual mannequin generated by AI •⁠ ⁠Enables voice-based interaction with an integrated conversational model •⁠ ⁠Detects the user’s face and gaze for contextual interaction •⁠ ⁠Displays essential daily information: time, weather, reminders, and notes

How we built it

•⁠ ⁠Computer Vision: Face detection and gaze tracking for user awareness •⁠ ⁠AI Models:

  • Diffusion model for generating outfit suggestions
  • Lightweight conversational AI (Mistral 7B) for voice interaction •⁠ ⁠Interface Layer: Smart display overlay integrated into a mirror surface •⁠ ⁠System Integration: Combined sensors, display, and backend AI pipelines into a seamless real-time experience

Challenges we ran into

•⁠ ⁠Achieving real-time performance with multiple AI components •⁠ ⁠Synchronizing face detection, gaze tracking, and UI responsiveness •⁠ ⁠Ensuring accurate outfit generation that feels personalized •⁠ ⁠Integrating voice interaction smoothly without latency issues

Accomplishments that we're proud of

•⁠ ⁠Successfully combining computer vision, generative AI, and voice AI in one system •⁠ ⁠Creating a natural, hands-free interaction experience •⁠ ⁠Building a visually appealing and functional smart interface •⁠ ⁠Delivering a working prototype that feels practical, not just conceptual

What we learned

•⁠ ⁠Multimodal systems (vision + voice + generative AI) require careful optimization •⁠ ⁠User experience matters as much as technical capability •⁠ ⁠Real-time AI integration is more complex than isolated model usage •⁠ ⁠Clear system architecture is critical when combining multiple AI pipelines

What's next for Smart Mirror

•⁠ ⁠Improve personalization using user preferences and history •⁠ ⁠Add real-time wardrobe integration (scan your clothes for suggestions) •⁠ ⁠Enhance voice assistant capabilities with deeper context awareness •⁠ ⁠Optimize hardware for faster and more efficient processing •⁠ ⁠Expand features like fitness tracking, skincare analysis, and smart home control

Share this project:

Updates