Research on the Estimation of Gaze Location for Head-eye Coordination Movement

Abstract Sight is the main source for humans to obtain information from outside world. Due structure of human eye [1], range sight limited. For this reason, people need constantly move their line when observing surrounding environment and target, movement based on coordinated head eye[2]. Therefore, key issue gaze research how correctly establish relationship between head-eye movement. Taking simulated flight as background, paper collects a large number images through designed “three-camera eight-light source” data acquisition platform, proposes estimation method combination appearance features, which effectively combines The coordination Then, ResNet-18 deep residual network traditional BP neural are used complete effective fusion pose features in process capturing so realize accurate drop point, its average accuracy up 89.9%.