UFII COVID-19 SEED Awardees Virtual Seminar Series
“Computer Scientist, Engineer, Political Scientist, and Two Criminologists Walk into a… Hazardous Behavior Data Collection Project?!” by Dr. Read Hayes
Research Scientist and Director, Loss Prevention Research Council
Also Presenting: Dr. Cory Lowe
Friday, October 30, 2020
Because of COVID-19, previously common behaviors such as clearing one’s throat in a public place can evoke fear in those around us; furthermore, there have been numerous reports of retail guests engaging in hazardous behaviors (e.g., coughing and spitting at or on people, checkouts, and merchandise) to intimidate others, engage in retail crimes, or avoid apprehension for committing crimes.
If hazardous coughing behaviors can be quickly detected, they can be addressed. Computer vision technologies provide a tremendous opportunity to rapidly detect these behaviors in public places; unfortunately, the annotated video clip data necessary to develop cough-detection algorithms are not readily available.
Therefore, our team has designed a project to develop a publicly available collection of coughing videos that can be: (1) used in a cough-detection algorithm contest sponsored by loss prevention industry partners, and (2) released to the broader community of individuals working in AI and computer vision to further research in this area. This presentation reviews our novel, interdisciplinary project and the joys, challenges, and ethical considerations associated with it so far