english  | Home | Impressum | Datenschutz | KIT

Semantic Segmentation of Objects Close to the User for Improved Perception in Augmented Reality

Semantic Segmentation of Objects Close to the User for Improved Perception in Augmented Reality
Typ:Bachelor-/ Masterarbeit
Datum:vergeben
Betreuer:

M. Sc. Simon Stock

Semantic Segmentation of Objects Close to the User for Improved Perception in Augmented Reality

.
.

Background

The human eye is one of the most complex systems of the human body. The sense of seeing is the only way to perceive our environment over distance at a high resolution and many people will agree that it is the most important human sense.
With increasing age the possibility of having a vision disorder increases. Also the near point accommodation performance in diopters decreases with age. At some point almost everyone develops Presbyopia. This can be correct with ‘reading glasses’. However there are age related diseases which are not easily corrected with glasses, these include: Cataract, Glaucoma and Age Related Macular Disease (AMD). Virtual Reality (VR) and Augmented Reality (AR) might can be used to improve the perception quality of affected people. Furthermore an improved visual perception might also be achievable for healthy eyes with the means of modern technology. 

 

Tasks

  • Semantic segmentation of the near field
  • Recognition of e.g.: Smartphones, Flyers, Books, Newspaper, Cards …
  • For the use in Augmented Reality
  • Real Time Processing
  • Integration with the Windows Mixed Reality platform and HoloLens 

 

Required skills

  • Motivation and creativity!
  • Programming skills
  • Computer Vision or Machine Learning experience is a plus