A Solution for the Occlusion Problem Encountered in the Virtual Reality Based Training Application for Cardiopulmonary Resuscitation

Abstract

In the field of computer vision, occlusion is a fundamental problem which is the effect of one object blocking another object from view. Though significant progress has been achieved in the object detection tasks, it is still challenging to detect the objects in the presence of occlusion. In the use of Virtual Reality (VR) based training application for Cardiopulmonary Resuscitation (CPR), when one hand crosses over the other at the posture of performing chest compressions, the lower hand gets occluded by the other. This study proposes a solution for the aforementioned problem of hand occlusion encountered in the implementation on VR based application to train emergency medical personnel in CPR. Real time hand detection improves the realism of the application, because hands are the administering organ of the user in performing chest compressions. The study has utilized Teensy 3.2 USB development board to build a wearable device, which provides a solution to the difficulties faced by the user whilst performing chest compressions to the patient in a VR environment. The inherent limitations of the devices used in this study have to be overcome using manipulations at the code level of the application. The outcome of this research has been succesfull in hand tracking even in the presence of occlusion. This solution has been well recieved by the most of all the users who have participated in the evaluation. Thereby, the solution provided in this study has enhanced the realism of the VR based CPR training application.