Deep Learning-Based Autonomous Driving System with OpenCV Integration

This paper presents the design and development of a cost-efficient, real-time autonomous driving prototype that leverages Raspberry Pi and deep learning techniques for intelligent navigation. The proposed system integrates the YOLOv5 object detection framework with OpenCV-based lane detection and a lightweight CNN for traffic light classification. An ultrasonic sensor module is used for obstacle proximity awareness, and all modules are combined into a unified architecture optimized for execution on embedded hardware. A Streamlit-based dashboard provides interactive feedback and monitoring. The system demonstrates strong performance in terms of detection accuracy and response time, validating its potential for use in low-cost driver assistance applications and retrofitting older vehicles.