Boosted Gaze Gesture Recognition Using Underlying Head Orientation Sequence

People find it challenging to control smart systems with complex gaze gestures due the vulnerability of eye saccades. Instead, existing works achieved good recognition accuracy simple because sufficient points but have limited applications compared gestures. Complex need a composition multiple subunits fixation contain sequence that are clustered and rotated an underlying head orientation relationship. This paper proposes new set angles as sequences recognize Eye powerful influence on gesture formation. The was obtained by aligning moving average (SMA) denoise interpolate gap between successive fixations. aligned utilized train sequential machine learning (ML) algorithms. To evaluate performance proposed method, we recruited recorded features ten participants using tracker. results show Boosted Hidden Markov Models (HMM) Random Subspace methods best accuracies 94.72% 98.1% for complex, respectively, which outperformed conventional methods.