Final Year Project 2026
FraudShield
AI-powered financial fraud detection for Windows. Six machine learning models analyse every transaction in real time — catching fraud before it costs you.
AI-powered financial fraud detection for Windows. Six machine learning models analyse every transaction in real time — catching fraud before it costs you.
Ensemble of six trained models — Logistic Regression, Random Forest, Decision Tree, SVM, Isolation Forest, and K-Means — vote on every transaction.
Instant fraud alerts with Low, Medium, and High severity levels, sound notifications, and a full resolution audit trail.
Live stat cards, fraud trend charts, risk distribution histograms, and period-over-period comparisons — all in one view.
bcrypt password hashing, Fernet-encrypted account numbers, session timeout, and account lockout after failed attempts.
Upload tens of thousands of transactions at once. Supports 100,000+ rows with background processing so the UI stays responsive.
Full user management, AI threshold controls (60–90%), system activity logs, and one-click database backup and restore.
Add transactions manually or bulk-upload a CSV file. Account numbers are encrypted automatically.
The ensemble model scores every transaction 0–100. Anything above your configured threshold is flagged as fraud.
Review flagged transactions, resolve alerts, and export professional PDF or Excel reports with date filters.
Free Windows installer. No Python. No configuration. Just download and run.
⬇ Download FraudShield v1.0.0Windows 10 / 11 · 64-bit · 2 GB disk space · Admin rights required for install
Project Supervisor
University of Central Punjab, Lahore
AI Model Development
& Backend Systems
Database Design
& Desktop UI
Reporting, Integration
& Deployment