Horologium: Timeless Commerce

• 

  Long screenshot of website

🕰️ Inspiration The love for timeless craftsmanship and the elegance of a watch inspired Horologium. We wanted to create a digital space where watch enthusiasts could explore, try on, and buy watches with ease and confidence.

Traditional watch-selling platforms lack personalization and interactive experience. We envisioned Horologium to combine e-commerce with modern AI and AR technologies to bridge that gap.

🔧 What We Learned How to integrate AR for virtual try-on using web-based camera access

Implementing AI recommendation engines using user preferences and history

Building scalable, responsive UIs with modern frameworks

Managing real-time product data updates and pricing APIs

⚙️ How We Built It Frontend: Built with Next.js and Tailwind CSS, optimized for performance and responsiveness

Backend: Powered by FastAPI, handling all API routes and watch catalog data

AI Module: Trained recommendation system using scikit-learn and deployed it with ONNX

AR Try-on: Implemented using TensorFlow.js and MediaPipe for hand/face detection

Database: MongoDB Atlas for fast, flexible document storage

Hosting: Deployed on Vercel (frontend) and Render (backend)

🚧 Challenges Getting virtual try-on to align watches accurately on the wrist using a webcam

Optimizing the AI model for real-time recommendation without delay

Managing dynamic data sync between backend and UI

Time constraints during integration and deployment in a hackathon setting

Built With

• bolt
• react
• supabase
• vite