Mixed Reality with Hardware Acceleration: implementing the multimodal user interface vision

doi:10.1109/AIVR56993.2022.00031

Mixed Reality with Hardware Acceleration: implementing the multimodal user interface vision

2022-12-01

0

PRE

AI

N

Nektarios Deligiannakis *

M

Maria-Evangelia Pavlopoulou

V

Vassilis Papataxiarxis

S

Stathes Hadjiefthymiades

DOI：10.1109/AIVR56993.2022.00031

原文链接

原文求助

摘要

En 中文

A Mixed Reality prototype application is proposed using the HoloLens Head Mounted Display device delegating AI tasks to edge or cloud systems. Processing and Battery life limitations of Head Mounted Display (HMD) devices do not allow these tasks to be executed for substantial periods of time without having an assistive external infrastructure to computationally support this process. Additionally, persistent and time sensitive AI procedures are delegated to achieve reliable and robust results. The architecture delegates the detection process to NVIDIA Jetson TX2, a GPU-accelerated, embedded and power-efficient device. We apply YOLO as the deep learning algorithm for object detection and use trained models on the Microsoft's Common Objects in Context (COCO) dataset. We project visual and auditory information about the discovered objects.

Keyword:

Mixed Reality

Object Detection

Head Mounted Display

Artificial Intelligence

Mixed Reality with Hardware Acceleration: implementing the multimodal user interface vision

摘要

期刊

机构

引用论文