Antique Analyzer is a project created for the "Projekt zespołowy" (Team Project) course by a team of Computer Science Master's students at the University of Information Technology and Management in Rzeszów (UITM).

The platform addresses the problem of counterfeit antiques in e-commerce. It combines Artificial Intelligence (multimodal analysis of images and text) with Community Verification (experts and users) to assess the authenticity of items listed on platforms like Allegro, OLX, and eBay.

• 🔍 AI-Powered Verification: Uses EfficientNet-B0 (Vision) and DistilBERT (NLP) to analyze auction photos and descriptions, detecting potential scams or replicas.
• 🕷️ Automated Scraping: Intelligent web scrapers (Selenium, Apify) automatically fetch auction data from URL links.
• 👥 Expert Crowdsourcing: A dedicated panel for domain experts (Numismatics, Furniture, etc.) to validate AI results and provide professional opinions.
• 🗳️ Community Voting: A democratic voting system where the community can agree or disagree with the verdict.
• 📊 Confidence Scoring: The system provides an authenticity probability score (Original vs. Scam/Replica).

The system operates on a microservices architecture:

1. Core Backend (Ruby on Rails): Manages users, auth, database, auctions, and community interactions.
2. AI Service (Python/FastAPI): Exposes endpoints for scraping and running the ML inference.
3. Frontend (React/TypeScript): A modern, responsive web interface.

Here are the open-source components of our platform:

We are a team of Master's students passionate about AI and Data Science.

Kamil Kukiełka Leader & Data Scientist

Bartosz Kawa Backend Developer

Michał Żychowski Data Engineer

Kamil Michna Frontend Developer