Real-Time Object Detection for an Intelligent Retail Checkout System

Customers are made to wait for extended hours in busy shopping settings as traditional limits of barcode and RFID-based billing systems have a predisposition towards errors and inefficiencies in scanning untagged or weighted items. The methods that are in use today can't keep up with the demand and lead to queues at check-out counters causing customer discontentment. This paper presents an idea of a novel Smart Billing Counter, based on Object Detection using advanced deep learning, YOLOv8, and OCR technologies for smooth billing processes. It captures the image of items placed at the counter, identifies what it is with minimal error, extracts necessary information from labels concerning weighted items, and generates a bill automatically. The unique capability of this solution is to recognize identical items in one frame with total cost by quantity to optimize checkout times and minimize the possibility of errors or fraud. At an inference speed of 37.2ms per image with 93.6% accuracy, this significantly enhances operational efficiency and ensures that grocery retailers provide their customers with smooth, error-free, and swift checkout processes. This innovation is likely, therefore, to bring about a transformation in retail billing systems. It will enhance customer satisfaction and operations efficiency quite highly.