25th International Conference on Pattern Recognition Workshops, ICPR 2020, 10/01/2021-15/01/2021,Lecture notes in computer science 12662 LNCS, pp. 447–459

Anno: 2021

Tipologia: contributo-in-atti-di-convegno

Autori: Milella A.; Marani R.; Petitti A.; Cicirelli G.; D'orazio T.

Keywords: 3D reconstruction and modelling,Intelligent retail,RGB-D sensors

This paper presents a 3D Vision-Based Shelf Monitoring system (3D-VSM) aimed at automatically estimating the On-Shelf Availability (OSA) of products in a retail store. The proposed solution exploits 3D data returned by a consumer-grade depth sensor to provide up-to-date information about product availability for customer purchase and eventually generate alerts on Out-Of-Stock (OOS) events, based on the comparison between a reference model of the shelf and its current status. The main advantage is that no a priori knowledge about the product characteristics is required, while the shelf reference model is automatically built, based on an initial training stage. The 3D-VSM system is integrated into an e-commerce application for electronic shopping and home delivery, developed in the context of the E-SHELF research project. Experimental tests carried out in a retail store show that the system is able to accurately estimate the on-shelf availability of products, overcoming time and labour problems of conventional audits.