●Technical introduction:
1. By applying the developed YOLOv3-tiny-based model to pupil tracking performance test, the detection accuracy reaches 80％, and the recall rate is close to 83％.
2. The average visible-light pupil tracking errors are smaller than 2 pixels for the training mode and 5 pixels for the cross-person test.
3. After calibration, the average gaze tracking errors are smaller than 2.9 and 3.5 degrees at the training and testing modes, respectively.
4. The proposed visible-light wearable gaze tracking system performs up to 20 frames per second (FPS) on the GPU-based software embedded platform.

●The scientific breakthrough of technology:
1. Based on a nearfield visible-light eye image dataset, the YOLOv3-tiny based deep-learning model achieves real-time and accurate detection of the pupil position at the visible-light mode.
2. The proposed design detects the position of the pupil＇s object at any eyeball movement conditions, outperforms traditional designs without deep-learning technologies.
3. The proposed pupil tracking technology overcomes efficiently the light and shadow interference at the near-eye visible-light mode, and the detection accuracy of a pupil’s location is higher than the conventional approaches.

●Industrial applicability of technology:
Wearable eye tracker can track and measure eyeball position and eye movement information. It has been widely used in visual system, psychology, and cognitive linguistics. In education and learning research, it can analyze the eye movement of each tester to support a unique teaching method. In market research and consumer surveys, eye trackers can be used to learn which products and packaging designs are more attractive to consumers. The developed scheme can also be applied to various domains such as driving safety monitoring and human-computer interaction/ interface designs.

線上展網址：
https://tievirtual.twtm.com.tw/iframe/bd77728d-dc93-4ef2-8b2e-2c348882d6cc?group=23bfb1fa-dd5b-4836-81a1-4a1809b1bae5&lang=en