Convolutional neural network implementation for eye movement recognition based on video

The states and movements of human eyes contain a lot of useful information, and these provide an attractive alternative plan to the cumbersome interface devices for human-computer interaction (HCI). As a result, the research on recognition of unit eye movement has become a hotspot in human activity recognition. In this paper, we proposed an eye movement recognition method based on convolutional neural network (CNN). An image dataset with eye movement was built for training. We conducted the experiment by training 16000 eye movement images. The experimental results showed that the highest accuracy achieved 99.7062% by using 16 kernels of size 7 × 7 in the first convolutional layer and 16 kernels of size 7 × 7 in second. Through the comparison experiment, it has been turned out that recognition rate of CNN was higher than using support vector machine (SVM), back propagation neural network (BP) and eye movement recognition based on electrooculography (EMR-EOG).