On high-precision chessboard detection on static scene videos from mobile eye-tracking devices

Eye-tracking analysis require annotation of a scene video by information on the target of the gaze. We develop a technique for automatic high-precision scene annotation for mobile head-mounted eye-trackers. The solution combines computer-vision techniques for scene recognition, 3-D modeling of the recognized objects, and head movement compensation using gyroscope and accelerometer information provided by the eye-tracking device. In this paper we address the problem of recognition of a chessboard, while the approach may be applied to other situations with static scenes.