Current trends and future directions in eye tracking technology: A literature review

In recent years, literature on eye tracking has seen a significant rise in interest. While previous approaches were often highly intrusive, modern solutions mostly rely on non-intrusive devices such as glasses, headsets, or table cameras. These systems are increasingly reliant on self-assessing eye tracking algorithms powered by advanced machine learning techniques, and for this reason it is fundamental to identify optimized approaches which can work with good performances and reduced computational times. This work serves as a complete review of the current state of the art in available equipment and algorithms, introducing a novel taxonomy of approaches, with a particular emphasis on machine learning and neural networks. Comparative tables of the reviewed studies will also be provided, with an in-depth analysis of the current best algorithms. Additionally, the work will explore the potential applications of eye tracking in everyday life, with a focus on the medical and psychological fields, highlighting its use for both diagnostic and therapeutic purposes. A discussion of the current limitations will be provided to identify the shortcomings of state of the art in eye tracking and pinpoint the aspects which need improvements in future research. Finally, the discussion will address possible future advancements in eye tracking, particularly regarding datasets and algorithmic improvements, providing some directions.