Our architecture estimates eye region landmarks with a stacked-hourglass network trained on synthetic data (UnityEyes), evaluating directly on eye images taken in unconstrained real-world settings. The landmark coordinates can directly be used for model or feature-based gaze estimation.
Conventional feature-based and model-based gaze estimation methods have proven to perform well in settings with controlled illumination and specialized cameras. In unconstrained real-world settings, however, such methods are surpassed by recent appearance-based methods due to difficulties in modeling factors such as illumination changes and other visual artifacts. We present a novel learning-based method for eye region landmark localization that enables conventional methods to be competitive to latest appearance-based methods. Despite having been trained exclusively on synthetic data, our method exceeds the state of the art for iris localization and eye shape registration on real-world imagery. We then use the detected landmarks as input to iterative model-fitting and lightweight learning-based gaze estimation methods. Our approach outperforms existing model-fitting and appearance-based methods in the context of person-independent and personalized gaze estimation.
We would like to thank Erroll Wood, Tadas Baltrusaitis, and Wolfgang Fuhl for their help. This work was supported in part by ERC Grant OPTINT (StG-2016-717054), the Cluster of Excellence on Multimodal Computing and Interaction at Saarland University, Germany, and a JST CREST research grant (JPMJCR14E1), Japan.