SpendSmart is a modern, automated expense tracking pipeline that eliminates manual data entry. It uses Computer Vision (LLMs) to extract data from receipt images. The data is visualized in a real-time Streamlit Dashboard.
Screencast from 2026-01-27 15-59-13.webm Screencast from 2026-01-27 16-00-05.webm
- 🧾 AI Receipt Scanning: Upload a photo of any receipt; the system extracts the Merchant, Date, Total Amount, Currency, and Category automatically using GPT-4o / Gemini models.
- 📊 Interactive Dashboard: Analyze spending habits with Plotly charts (Donut charts, Trend lines) and KPI cards.
- ✍️ Smart Editing: A dedicated "AI Review" sidebar allows you to correct AI mistakes before finalizing the data.
- 🗄️ Backend: Data is stored in a structured PostgreSQL database using SQLAlchemy ORM.
- Frontend: Streamlit, Plotly Express
- Backend: Python 3.12
- Database: PostgreSQL (with
psycopg2&SQLAlchemy) - AI/LLM: OpenRouter API (Accessing GPT-4o-mini / Gemini 2.0 Flash)
- Image Processing: Pillow (PIL)
git clone [https://github.com/yourusername/SpendSmart.git](https://github.com/yourusername/SpendSmart.git)
cd SpendSmart