The ATM Fraud Detection System is an intelligent security solution designed to detect and prevent fraudulent activities at Automated Teller Machines using machine learning and deep learning techniques. The system integrates real-time video surveillance with transaction and user behavior analysis to identify suspicious activities effectively. Initially, machine learning models are trained using historical datasets. Live video footage captured from ATM cameras is continuously analyzed to detect theft-related activities, and if suspicious behavior is identified, encrypted video footage along with alert messages and emails are sent to nearby police stations and security officials. The system also monitors transaction patterns to detect irregular cash withdrawals, such as multiple failed PIN attempts, unusually large withdrawals exceeding predefined limits, repeated high-value transactions, or transactions not completed within a specific time frame. User behavior profiling is performed by analyzing historical transaction data to identify deviations, abnormal transaction sequences, unusual spending patterns, and transactions at odd hours or distant locations. To ensure data security, all alerts, messages, and video footage are transmitted using encrypted communication channels, employing AES encryption for video data and RSA encryption for text-based alerts. Additionally, deep learning-based card authentication validates card details and matches live images with registered user data, enhancing overall ATM security and fraud prevention.

• Modern and responsive design
• Clean and maintainable code
• Full documentation included
• Ready to deploy