AI-based machine vision for retail self-checkout system
March 1, 2022
In recent years advances in computing power, research in AI algorithms and the availability of large datasets have enabled the rise of reliable object identification and tracking. This thesis describes the development and predictionmodel considerations of a object detection system for retail items. This project has been conducted as a collaboration between ETH Zürich and the start-up company AI Retailer Systems, who wants to automate parts of the retailing experience, namely the checkout procedure in a retail store. This is to be achieved by tracking and identifying objects that the customer puts in their physical shopping cart, and then using this information to build a “virtual shopping cart”, which would potentially remove the need for a cashier to scan the items and handle payments in a convenience store. Development steps and considerations, as well as model choice and data dependencies are discussed, concluding with a final recommendation for which object detection model should be deployed for the first prototype.
