Portion Size Estimation Performance when Using Resizable Visual Aids of Matching Shape and Type in Virtual Reality

In nutrition research, measuring dietary intake is an important step in gaining insight into how diet contributes to the prevalence of non-communicable diseases. It is crucial to estimate portion sizes accurately, but most people find this difficult. Methods based on providing 3D virtual visual aids in, for example, augmented reality (AR) have been developed to improve estimation accuracy, but it is unclear how to design applications in order to improve accuracy, for example, whether the visual aid should be a realistic solid food replica or a more basic geometric shape. In this paper, we provide highly-needed basic research insights: we compared the performance of two common visual aid approaches and of free estimation in virtual reality, to better understand human portion size estimation and determine which approach shows more potential for portion size estimation in AR. The results showed that using realistic resizable virtual food replicas resulted in greater performance in terms of accuracy and precision than when using a basic cube wireframe and when using free estimation. Future research on visual aids should look into further potential benefits of using resizable aids of equal shape and food type.