Inspiration
Imagine it’s Saturday morning, and you have your usual grocery list in hand. Like every week, you’ll hit Costco, Walmart, and Kroger, picking up essentials. But here’s the thing—let’s say you need tomatoes. They’re at each store, but how do you know which one has the best price?
This is where Paralavi steps in. Instead of guessing, Paralavi’s algorithm does the hard work for you. It searches for the best deals across your chosen stores, compares prices in real time, and even maps out the most efficient route, helping you save time and money effortlessly. With Paralavi, every shopping trip becomes a smart shopping strategy.
What it does
Paralavi is a receipt management and shopping optimization app that uses AI to recommend the best store deals and track spending. With OCR, it extracts details from scanned receipts, categorizes expenses, and suggests budget-friendly shopping options across stores like Costco, Kroger, and Walmart.
How we built it
Paralavi leverages:
- Intel’s OpenVINO for high-speed OCR, converting receipt images into structured data.
- IPEX for training a price optimization model that analyzes user shopping habits and price trends to recommend cost-saving shopping lists.
- InterSystems IRIS for data storage and vector searches, enabling quick retrieval of similar items and prices.
- Intel Tiber Developer Cloud Development for building and running the model on the IDC Jupyter Notebook.
Challenges we ran into
Our biggest challenge was fine-tuning the OCR model to accurately process varied receipt formats. Integrating InterSystems IRIS for real-time, similarity-based searches also required significant data structuring, which often deemed hard to do.
Accomplishments that we're proud of
- Enhanced OCR Performance: Achieved accurate receipt data extraction with Intel’s OpenVINO, making processing fast and reliable.
- Effective Price Optimization: Attempted a model that learns from historical data to recommend cost-effective shopping lists.
- User-Centric Design: Built an intuitive interface with real-time insights, making budgeting and shopping easier.
What we learned
- Hatim worked with vector-based searches in InterSystems IRIS, structuring data efficiently for fast, accurate retrieval.
- Adit coordinated the integration of various components to ensure proper design of the app.
- Sahi optimized OCR performance with Intel’s IPEX and OpenVINO, ensuring reliable data extraction from diverse receipts.
- Sherwin used machine learning to develop an effective price optimization model, learning to handle complex shopping patterns.
What's next for Paralavi
Our next steps include deploying Paralavi to the cloud for broader accessibility, adding more stores, and integrating budget alerts. We’re also planning to enhance the AI model for even more precise price predictions and personalized recommendations.
Log in or sign up for Devpost to join the conversation.