We develop a classifier model trained to analyze anonymized skeletal data of passers-by at interactive public displays to determine whether an interaction has occured. The test setup and data collection methods are described. The skeletal data is preprocessed to highlight more relevant bodyparts. The performance of the finished model will be evaluated statistically and compared to approaches using human observers from other research.
Lacher, J., Bieschke, L., Michalowski, F., & Münch, J. (2023). Using machine learning to determine attention towards public displays from skeletal data. In Mensch und Computer 2023 - Workshopband.