| || || Chaudhry, Shonal|
| || || Mobile based face recognition for visually impaired persons|
Institution: University of the South Pacific.
Subject: Human face recognition (Computer science), Computerized self-help devices for people with disabilities, People with visual disabilities
Call No.: pac TA 1650 .C43 2015
Copyright:Under 10% of this thesis may be copied without the authors written permission
Abstract: One of the most difficult tasks faced by visually impaired persons is identification of people. While naturally, voice recognition is a common method of identification, it is an intuitive and difficult process. With the availability of mobile devices, these people can be assisted through an additional method of identification through intelligent software based on computer vision techniques. This thesis presents the design and implementation of a face detection and recognition system for the visually impaired through the use of mobile devices. It consists of a mobile device, mobile application and an online server-based support system. Experiments are performed on the mobile application to determine the detection and recognition performance. They are tested on a custom video dataset created for analysing the performance of the program in different situations which include lighting conditions, scale, motion blur and obstructions. Computer vision methods such as convolutional neural networks and cascade classifiers are used to further investigate performance. Experiment results show good performance with an accuracy of upto 94% in conditions where adequate lighting is present and lower performance in poorly-lit areas.