Eye centre localisation with convolutional neural networks in high- and low-resolution images

Zhang, Wenhao; Smith, Melvyn L.

Abstract

Eye centre localisation is critical to eye tracking systems of various forms and with applications in variety of disciplines. An active eye tracking approach can achieve a high accuracy by leveraging active illumination to gain an enhanced contrast of the pupil to its neighbourhood area. While this approach is commonly adopted by commercial eye trackers, a dependency on IR lights can drastically increase system complexity and cost, and can limit its range of tracking, while reducing system usability. This paper investigates into a passive eye centre localisation approach, based on a single camera, utilising convolutional neural networks. A number of model architectures were experimented with, including the Inception-v3, NASNet, MobileNetV2, and EfficientNetV2. An accuracy of 99.34% with a 0.05 normalised error was achieved on the BioID dataset, which outperformed four other state-of-the-art methods in comparison. A means to further improve this performance on high-resolution data was proposed; and it was validated on a high-resolution dataset containing 12,381 one-megapixel images. When assessed in a typical eye tracking scenario, an average eye tracking error of 0.87% was reported, comparable to that of a much more expensive commercial eye tracker.

• Journal Article Type: Article
• Acceptance Date: May 10, 2022
• Online Publication Date: Jul 15, 2022
• Publication Date: Jan 1, 2022
• Deposit Date: Jul 19, 2022
• Publicly Available Date: Jan 2, 2023
• Journal: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
• Print ISSN: 0302-9743
• Publisher: Springer Verlag
• Volume: 13375 LNCS
• Pages: 373-384
• Series Title: Lecture Notes in Computer Science
• ISBN: 9783031105210; 9783031105227
• DOI: https://doi.org/10.1007/978-3-031-10522-7_26
• Keywords: eye centre localisation; eye tracking; deep learning
• Public URL: https://uwe-repository.worktribe.com/output/9711805
• Publisher URL: https://link.springer.com/chapter/10.1007/978-3-031-10522-7_26
• Additional Information: First Online: 15 July 2022; Conference Acronym: ICCSA; Conference Name: International Conference on Computational Science and Its Applications; Conference City: Malaga; Conference Country: Spain; Conference Year: 2022; Conference Start Date: 4 July 2022; Conference End Date: 7 July 2022; Conference ID: iccsa2022; Conference URL: https://iccsa.org/; Type: Single-blind; Conference Management System: CyberChair 4; Number of Submissions Sent for Review: 279; Number of Full Papers Accepted: 57; Number of Short Papers Accepted: 24; Acceptance Rate of Full Papers: 20% - The value is computed by the equation "Number of Full Papers Accepted / Number of Submissions Sent for Review * 100" and then rounded to a whole number.; Average Number of Reviews per Paper: 2.6; Average Number of Papers per Reviewer: 8.7; External Reviewers Involved: Yes; Additional Info on Review Process: 285 Workshop submission accepted out of 815 submissions