Smart Shopping Cart

Abstract

This paper consists of developing a hardware and software implementation of a smart shopping cart, focused on modularity, low price and simple use. This solution can be modified and the usage can be started shortly by traders. It tends to meet as many of the end-user needs as possible.The system uses a mechanism to recognize products that are inserted into the cart using detection with a camera and a pre-trained neural network, the inference of which is running on a specific Intel calculation module, without the need for network connectivity. The project also presents a method of training and using a neural network for object detection, that runs with the Intel Neural Compute Stick. At the same time, in the evaluation of the results is demonstrated the increased performance against inference using a CPU, along with the comparison of versions 1 and 2 of the neural stick.