Publish In
International Journal of Advances in Electronics and Computer Science-IJAECS
Journal Home
Volume Issue
Volume-7,Issue-5  ( May, 2020 )
Paper Title
Improving Blink Detection Using Convolutional Neural Networks
Author Name
Thara Angskun, Wanchatpookhuntod, Jitimon Angskun
school of Information Technology, Suaranaree University of Technology, Thailand Digitech, Suranaree University of Technology, Thailand
Recently, million people are suffered from computer vision syndrome (a.k.a. CVS) because they are inappropriately using device screens. To prevent CVS, a software module called EyeGuard has been developed. This paper proposes an improved version of blink detection mechanism in the EyeGuard. The blink detection mechanism is implemented using convolution neural networks. The experimental evaluation reveals that the introduced mechanism has a mean absolute percentage error (MAPE) of 5.2 per cent and 4.8 per cent at the distance of 60 cm. and 80 cm., respectively. These values are significantly reduced from the original version which hasan MAPE of 14.25 per cent and 18.25 per centat the distance of 60 cm. and 80 cm., respectively. Keywords - Blink Detection, Computer Vision Syndrome, Convolutional Neural Networks.
  View Paper