Video-assisted Dentistry with Deep Neural Networks

D. Węsierski (Gdansk Univ. of Techn., Poland)

This study aims to develop a novel video-based technology to support the next generation of digital dentistry. By enabling enhanced and continuous intraoral visualization, the system has the potential to improve educational environments, clinical training, and real-time operational workflows in conservative dental procedures. The proposed approach is grounded in multi-task learning, where a unified neural network is trained to simultaneously perform multiple vision based tasks for enhanced human-system interaction. To facilitate this development, the study leverages existing and newly curated dental video datasets, including annotated sequences for training and evaluation.