Evaluation of state-of-The-Art pupil detection algorithms on remote eye images

Eye movements are a powerful source of information a well as the most intuitive form of interaction. Although eyetrackin technology is still in its infancy, it offers the greates potential for novel communication solutions and applications Whereas head-mounted eye-Trackers are widely use in research, several applications require most unintrusiv eye tracking, ideally realized by means of a single, low-cos camera placed away from the subject. However, such remot devices usually provide low resolution images an pose several challenges to gaze position estimation. Th key challenge in such a scenario is the robust detectio of the pupil center in the recorded image. We evaluate eight state-of-The-Art algorithms for pupil detection on thre manually labeled data sets recorded in remote tracking scenarios Among the evaluated algorithms, ElSe [6] prove to be the best performing approach on overall 3202 image from remote eye tracking, which include changing illumination occlusion, head movements, and off-Axial camer position. In addition, we contribute a new data set with 44 annotated images, recorded in a fixed setup with a low cos camera capable of using natural and infrared light.