The overall goal of this work is to validate a low vision aid system that uses gaze as a pointing tool and provides smart magnification. We conclude that smart visual enhancement techniques as well as gaze contingency should improve the efficiency of assistive technology for the visually impaired.
A low vision aid, using gaze-contingent visual enhancement and primarily intended to help reading with central vision loss, was recently designed and tested with simulated scotoma. Here, we present a validation of this system for face recognition in age-related macular degeneration patients.
Twelve individuals with binocular central vision loss were recruited and tested on a face identification–matching task. Gaze position was measured in real time, thanks to an eye tracker. In the visual enhancement condition, at any time during the screen exploration, the fixated face was segregated from background and considered as a region of interest that could be magnified into a region of augmented vision by the participant, if desired. In the natural exploration condition, participants also performed the matching task but without the visual aid. Response time and accuracy were analyzed with mixed-effects models to (1) compare the performance with and without visual aid and (2) estimate the usability of the system.
On average, the percentage of correct response for the natural exploration condition was 41%. This value was significantly increased to 63% with visual enhancement (95% confidence interval, 45 to 78%). For the large majority of our participants (83%), this improvement was accompanied by moderate increase in response time, suggesting a real functional benefit for these individuals.
Without visual enhancement, participants with age-related macular degeneration performed poorly, confirming their struggle for face recognition and the need to use efficient visual aids. Our system significantly improved face identification accuracy by 55%, proving to be helpful under laboratory conditions.
1Laboratoire de Psychologie Cognitive, Aix-Marseille University, Marseille, France
2Laboratoire Bases, Corpus, Langage, Nice Sophia-Antipolis University, Nice, France
3Low Vision Clinic, University Hospital of La Timone, Marseille, France
4Department of Ophthalmology, University Hospital of La Timone, Aix-Marseille University, Marseille, France *email@example.com
Submitted: December 18, 2017
Accepted: May 22, 2018
Funding/Support: Essilor International (PhD grant; to CA) and Fondation de France (Berthe Fouassier post-doctoral grant; to AC).
Conflict of Interest Disclosure: None of the authors have reported a financial conflict of interest.
Author Contributions: Conceptualization: AC, CA, EC; Data Curation: AC, CA; Formal Analysis: AC; Funding Acquisition: EC; Investigation: CA, GF, EC; Methodology: AC, CA, EC; Project Administration: FM, LH, EC; Resources: GF, FM, LH, EC; Software: CA; Supervision: EC; Writing – Original Draft: AC; Writing – Review & Editing: AC, EC.