Area4U: Predicting Interaction Hotspots for Large Public Displays

Effective interaction with large public interactive displays (LPIDs) remains a significant challenge in human-computer interaction. This paper introduces "Area4U", a user interaction area prediction framework designed for enhanced interaction with LPIDs. Area4U employs lightweight deep-learning techniques to analyze webcam imagery and deduce potential users' location and motion information, utilizing this information to predict and allocate sub-regions users prefer for their usage, the framework streamlines the transition of potential users to active participants. Our model was trained on video data containing 212 sub-region selections from 12 participants. Results indicate that our method effectively identifies areas of interest to potential users in an experimental setting, proving the feasibility and effectiveness of our framework.