Phosphene Vision and AI: Representing the World

Keywords

Loading...
Thumbnail Image

Authors

Issue Date

2019-07-01

Language

en

Document type

Journal Title

Journal ISSN

Volume Title

Publisher

Title

ISSN

Volume

Issue

Startpage

Endpage

DOI

Abstract

Prosthetic vision relies on the observation that stimulating the visual cortex generates small visual percepts called phosphenes. Due to biological limitations, current cortical implants have low resolution. Researchers started investigating more complex image processing techniques in an attempt to circumvent these lim- itations and improve the quality of the phosphene image. Semantic segmentation has been proposed as a potential solution to this issue. However, these complex algorithms are computationally demanding. We investigated whether there are any advantages in using complex image processing techniques such as semantic segmen- tation over simpler methods, here represented by edge detection. We compared the two methods through a visual search task in simulated phosphene vision on healthy subjects, measuring reaction times as well as subjective opinions. Our re- sults show that participants performed better under the edge detection condition. The current state of the semantic segmentation model used in this work needs to be improved before it can yield better results than edge detection.

Description

Citation

Supervisor

Faculty

Faculteit der Sociale Wetenschappen