Programmable Errorless Face-Name Association Device with Real-Time Processing
Samuel Sisson, Kiran George · 2018
This paper presents a device to assist individuals with Alzheimer's Disease or Prosopagnosia disorder by performing facial recognition to provide face-name association training and facial-recognition capabilities. The device system is comprised of a Kangaroo Plus Mini PC, a Bluetooth earpiece, and a discreet wearable camera. The device will take a picture on request by its user and scans it for the faces of people it was taught to know. Upon detecting known faces, it gives out previously stored information about recognised individuals through an earpiece. The performance of the platform was tested to find the minimum required images per subject in the database and the delay incurred as a result. After testing, it was determined that five images per database subject with a resolution of 1280 x 720 pixels gave an accuracy of 99.333% at a range of eleven feet.