Object Recognition on Augmented Reality Glasses

Abstract

Employee training in fast-food restaurants is a long, practice-based process which is mainly done on the job. Employee performance during training directly affects service quality and customer satisfaction. In this study, it is aimed to optimize the training process in fast-food restaurants with use of augmented reality glasses. For this purpose, a comparative study is performed to determine the most suitable model of augmented reality glasses in the market. It is aimed to shorten the training period of the employee, to help the employee in this process and to introduce the objects around him. Light convolutional neural networks are compared to solve the object recognition problem on augmented reality glasses. As a result, MobileNet model is selected and fine-tuned to recognize the objects in a restaurant kitchen. The outcomes of this study will be used to fully train and supervise the employees without the need for a trainer in the future.