Inspiration
Managing personal finances can be overwhelming, especially when tracking expenses manually. We wanted to create a seamless solution that automates expense tracking by extracting data from receipts and categorizing spending intelligently. Inspired by the frustration of manual budgeting apps, we built MoneyMoves to leverage AI for effortless financial management, helping users stay on top of their spending without tedious data entry. With just a simple image file upload, we are able to automatically update the dashboard and other visual elements.
What it does
MoneyMoves simplifies expense tracking by automatically scanning receipts and categorizing your spending. When you upload a receipt, the app extracts the text, identifies the purchased items and total amount, and assigns them to budgeting categories like Groceries, Entertainment, or Personal Care. It then updates your financial dashboard in real time, showing how much you’ve spent compared to your monthly goals. By automating manual data entry, MoneyMoves helps users stay on top of their finances effortlessly.
How we built it
We built MoneyMoves using OpenCV and PyTesseract for receipt text extraction, DeepSeek's LLM API for expense categorization, and Firebase for secure data storage and authentication. The frontend is developed with React and Tailwind CSS, while a Flask backend handles data processing and integration between components. A lot of our initial planning and testing was inside Intel Tiber, playing around with parameters in the DeepSeek learning module and also the OpenVino learning module. This helped us drastically implement ideas and features in our application.
Challenges we ran into
The biggest challenge we faced was integrating the AI Python features with our backend. We had to learn about Flask endpoints and figure out how to connect the output to our backend and database. Another major hurdle was figuring out how to improve accuracy with both the image-to-text scraper and the LLM classification.
Accomplishments that we're proud of
We are super proud of successfully creating a fully functional pipeline that transforms receipt images into categorized financial data. Our team integrated four different technologies seamlessly to deliver a cohesive product. This was a lot of our first time with these technologies, so this was a huge confidence boost that we were able to create a full product. Most importantly, we built something that genuinely solves a common pain point in many people's lives.
What we learned
This project taught us how computer vision and large language models can work together to solve real-world problems. We gained a deeper understanding of prompt engineering to improve the AI's categorization accuracy. The experience also highlighted the importance of thorough testing when integrating multiple systems. Lastly, we learned how to make neat and clean landing pages for our products.
What's next for MoneyMoves
The biggest thing we could've done in simply an hour more was to integrate more visual elements into our dashboard. We had already made graphs and pie charts that sync up with our database, but we didn't have enough time to integrate them into our application. Some long-term steps would be to perfect the features we have and publicly host our website. Over time, this has the potential to be a very powerful tool for anyone, and has monetization possibilities as well through subscription models.
Log in or sign up for Devpost to join the conversation.