📄 InvoiceShield

Invoice Anomaly Detection & Vendor Fraud Analysis System

InvoiceShield is an end-to-end system for processing invoices, detecting anomalies, and analyzing vendor risk using OCR, rule-based logic, and machine learning. It provides a usable dashboard for analytics, fraud scoring, and audit-ready reporting.

---

Key Features

• Invoice upload (PDF / PNG / JPG)
• OCR-based text extraction (EasyOCR)
• Rule-based + ML anomaly detection
• Vendor behavior clustering (DBSCAN)
• Fraud risk scoring with explanations
• Interactive Streamlit dashboard
• Downloadable PDF invoice reports
• FastAPI backend with Swagger UI

Approach

• OCR converts invoices into structured text
• Rules + ML models flag anomalous invoices
• DBSCAN clustering groups vendors by behavior and identifies outliers
• Fraud score (0–100) summarizes invoice risk with explainable reasons
• Results are stored and visualized through a dashboard

Project Structure

InvoiceShield/
├── src/
│   ├── api/
│   │   └── main.py
│   ├── ocr/
│   ├── parser/
│   ├── rules/
│   ├── fraud/
│   ├── reports/
│   └── db/
├── dashboard/
│   └── app.py
├── invoices.db
├── requirements.txt
└── README.md