Object Tracking in Augmented Reality: Enhancement Using Convolutional Neural Networks

Abstract

Augmented reality (AR) has been applied in maintenance, simulation, remote assistance, and other fields. One of the issues arising from these applications is how objects can be placed in physical environments using an AR system. Object placement consists of two processes: object detection and segmentation. Due to the importance of placement, in this paper, we propose using deep learning to address issues with the placement of objects through detection and segmentation in AR. Deep learning can help complete tasks by providing correct information about environmental changes in real-time situations. The problem is that it is rarely used in AR, which suggests a combination of deep learning-based object detection and instance segmentation with wearable AR technology to improve performance on complex tasks. In our work, we propose to address this problem by applying a convolutional neural network to facilitate the detection and segmentation of objects in real environments. To measure AR performance, we examined detection accuracy in environments with different intensities. The results of the experiment demonstrate satisfactory performance, reaching 98% for segmentation and accurate detection.