Comparative performance characterization of mobile AR frameworks in the context of AR-based grocery shopping applications

Abstract

A number of Augmented Reality (AR) frameworks are now available and used to support the development of mobile AR applications. In this paper, we measure and compare the recognition performance of the commercial AR frameworks and identify potential issues that can occur in the real application environment. For experiments, we assume a situation in which a consumer purchases food products in a grocery store and consider an application scenario in which AR content related to the products is displayed on a smartphone screen by recognizing such products. We use four performance metrics to compare the performance of the selected AR frameworks, Vuforia, ARCore, and MAXST. Experimental results show that Vuforia is relatively superior to the others. The limitation of the AR frameworks is also identified when they are used in a real grocery store environment.