EventDetection Use CNN to extract features from the eye images and feed those features into parallel LSTMs to do Gaze-In-world classification.