Faceboard: Hands-free Text-Entry based on head movement

Abstract

The introduction of smartphones allows billions of people to enter text on the go. However, users with limited hand functionality face challenges when entering text, as current solutions usually require additional hardware such as eye-tracking devices. This study explores a novel and hands-free input method for smartphones utilizing the camera of the smartphone. By using machine learning algorithms, head movements and blinking are processed and mapped onto the screen in real-time to allow swipe based input without the need to touch the phones’ screen.Evaluated by 16 participants, the average user managed to input text with 5.98 words per minute and a keystrokes per character value of 1.67. Other text-entry Methode like EyeSwipe proposed by Kurauchi et al. or Tiltwriter by Castellucci et al. achieved average speeds of 11.7 and 12.1 words per minute, which indicates room for improvement.Qualitative feedback was collected via System Usability Scale (SUS) and NASA-TLX, Faceboard showed an average SUS score of 59.1 and an average weighted TLX workload of 58. Compared to the generally accepted values of 68 for SUS and an average weighted workload of 42 for TLX, usability and workload were problems for participants. Participants showed the need for improvement in reducing physical and mental strain. Possible future work aims to enhance usability by making parameters configurable and introducing dwell-time selection instead of blinking.Keywords: Mobile Devices, Input Techniques, Accessibility, Head-based-Pointing, Human-Centered Design