The School of Computer Science is celebrating the achievement of master of science student Hai Ly, whose research paper on skeleton-based keystroke recognition in virtual reality received the Best Short Paper Award at the IEEE AIxVR 2025 Conference.
Co-authored with his supervisor, Dan Wu, the paper, "Towards Skeleton-Based Keystroke Recognition in Virtual Reality," introduces an innovative method that applies deep learning models to hand skeletons to improve text input in VR environments.
Professor Dan Wu and master of science student Hai Ly celebrate the latter's award for Best Short Paper at the IEEE AIxVR 2025 Conference.
The research aims to enhance the typing experience on mid-air virtual keyboards by leveraging skeleton-based deep learning techniques to predict keystrokes with high accuracy. Their proposed model achieved an impressive 94.61 per cent keystroke classification accuracy, outperforming existing methods and contributing to the development of more efficient VR-based office tools.
Recognizing the lack of publicly available datasets for keystroke recognition in VR, Ly and Dr. Wu also curated KeyVR, a new dataset designed to support further advancements in this field. The dataset is now accessible to the research community, promoting future innovation in virtual reality interaction technologies.
"Hai is a very hardworking student with exceptional skills in coding and writing," says Wu. "His dedication to solving real-world challenges in virtual reality interaction is truly commendable, and this recognition is well deserved."
The IEEE AIxVR Conference is an annual international event bringing together leading researchers and industry professionals in artificial intelligence, extended reality, and human-computer interaction. The 2025 conference ran Jan. 27 to 29 in Lisbon, Portugal.
