Customer detection and tracking using computer vision, YOLO, and hardware integration: a retail analytics approach

This paper presents a comprehensive system for customer detection and tracking in retail environments using computer vision technology, YOLOv4 object detection, and hardware integration. The system combines software components including OpenCV, Python programming, and YOLOv4 neural networks with hardware elements such as Arduino UNO, LCD displays, and custom PCB boards to provide real-time customer analytics. The implementation focuses on tracking customer movement within specific regions of interest (ROI) to help retail store owners analyse customer attraction performance and optimize store layouts. While the system successfully demonstrates the integration of computer vision with hardware components, performance limitations including low frame rates (1 FPS) and computational constraints highlight the need for GPU acceleration and more powerful hardware configurations. Recent advances in YOLO architectures and edge computing devices offer promising directions for improving system performance and practical deployment in retail environments.