Traditional retail shopping is a manual process in which customers select items, place them in a cart, and proceed to the checkout counter. Each product must then be scanned individually with a barcode reader, often resulting in long queues and delays during peak hours. To address this challenge, we propose a Smart Shopping Cart system that leverages computer vision and deep learning for automated product recognition and checkout. A custom YOLOv6 object detection model was trained on a curated, labeled product dataset to ensure accurate, real-time item recognition. The trained model was converted into ONNX format and seamlessly integrated into a web-based application. This application runs entirely on the edge, enabling in-browser inference for real-time detection without reliance on remote servers. The backend provides secure communication through JSON Web Token (JWT) authentication and connects to a MongoDB database to manage product details, user sessions, and purchase history. By automating manual tasks and streamlining the checkout process, the proposed system significantly enhances retail efficiency, offering an automated checkout solution accessible from any device equipped with a camera and Internet connection.