Eye State Detection Based on MTCNN and ESP

Eye state detection is an essential task in computer vision with diverse applications including emotion recognition, fatigue detection in high-risk areas, and computer interaction. This paper introduces a classification model called ERNet, which leverages the Efficient Spatial Pyramid (ESP) technique. The ERNet model is combined with the MTCNN to create a novel eye state detection model called MTED. The ERNet proposed in this paper has better performance than the existing model on the MRL Eyes 2018 dataset. Our ERNet model is more than 99.06% accurate on the MRL Eyes 2018 dataset while maintaining a high processing speed of 87 frames per second. Furthermore, the MTED model achieves a commendable accuracy of 95.41% on the ABD dataset.