Smart Trolley Application with an Approximate Multiplier Approach Using Embedded Platform

Abstract

Shopping malls, these days, are crowded and many people find it time-consuming and hard to shop. Locating a commodity coupled with standing in long queues for billing looks cumbersome. To overcome these difficulties, the authors of this paper, present the design of the product, smart trolley. When the customer enters the market, he has to log-in to the product and the map of the supermarket is displayed on the LCD screen. Later, once he starts shopping, upon entering each aisle, the details of the products in the aisle are displayed. For billing, the user has to scan the bar code for packed goods and based on the scanned code corresponding to the product, the price is displayed. For unpacked items such as fruits, vegetables etc, a picture of the commodity is taken which is classified with neural networks, the weight of the commodity is measured and the price is calculated with this information. Later, the customer has to click the finish button upon finishing shopping and pay via online transactions. This system design is validated with a prototype that has a camera, transducers, LCD screens RFID readers and raspberry Pi embedded in it. RFID technology is used for location tracking, cameras are used for scanning bar codes (packed goods) and taking snaps of commodities (unpacked) and a force transducer is used for measuring weight. Finally, the database images are multiplied with different level of approximate multipliers and various error level database has been created. The accuracy of all databases is tested with the VGG-16 network.

Deep learning for insight-based analytics to improve shopping experience

“…-VGGs: The VGG family, encompassing VGG-16 and VGG-19 [20], stands out as a renowned cornerstone utilized in computer vision and computer science tasks. The architecture of VGGs has demonstrated its effectiveness across numerous endeavors, including image classification, object detection, The authors Savio and Deepa [38] are proposed a Smart Trolley when is implemented a classifier a VGG-16 neural network gave an accuracy score of 98 percent and the error rate of the weight obtained using the load cell was less than 8 percent.…”