JewelryValuer

Inspiration

As someone who’s sold pre-owned luxury jewelry and watches, I saw a gap: slow, opaque valuations from jewelers, and confusing resale options. I wanted to make accurate, accessible valuations available to anyone—no appointments, no jargon, just a photo and instant insights.

What it does

JewelryValuer uses AI to analyze photos of jewelry/watches, detecting type, materials, brand, and condition. It delivers real-time resale/retail value ranges, personalized selling platform recommendations, and a purchasable insurance report for official documentation.

How we built it

We started with TensorFlow to train the AI model on thousands of jewelry/watch images for type/brand/material detection. The frontend (React) was built for intuitive photo upload/capture, while AWS hosted the backend (Python/FastAPI) to process analysis in under 10 seconds. Stripe API integrated payment for premium reports, and PDFKit auto-generates formatted insurance documents.

Challenges we ran into

Training the AI to accurately identify diamond characteristics (cut/color/clarity) from consumer photos was tough—lighting/angle distorted data. We also struggled with payment processing errors and ensuring PDF reports met insurance industry formatting standards for Western markets.

Accomplishments that we're proud of

The AI now completes analysis in <10 seconds with high accuracy for major luxury brands (Cartier, Rolex, Tiffany). We built a seamless free/premium tier model, and the insurance report is accepted by top Western insurance providers for jewelry coverage.

What we learned

I mastered balancing AI accuracy with user experience—consumers don’t need exact diamond measurements, just reliable ranges. I also learned the nuances of Western secondhand luxury markets, tailoring pricing logic to real transaction data from US/EU resale platforms.

What's next for this project

We’ll add integration with top Western selling platforms (eBay, The RealReal) to auto-generate listings, expand brand coverage to niche luxury makers, and build a mobile-optimized version for on-the-go valuations—all while refining AI for rare gemstone detection.

Built With

• amazon-web-services
• postgresql
• python
• react
• tensorflow