Surveillance VideoQuerying With A Human-in-the-Loop

SurvQ is a video monitoring system appropriate for surveillance applications such as those found in security and law enforcement. It performs real time object property identification and stores all data in a scalable DBMS. Standing queries implemented as database triggers are supported. SurvQ contains novel adaptive machine learning and algorithmic property classification. The application of SurvQ to assist the West Lafayette (IN) police department at identifying suspects in video is described. This paper also describes the basics of the SurvQ architecture and its human-in-the-loop interface designed to accelerate everyday police investigations. ACM Reference Format: Michael Stonebraker1, Bharat Bhargava2, Michael Cafarella3, Zachary Collins1, Jenna McClellan1, Aaron Sipser1, Tao Sun1, Alina Nesen2, KMA Solaiman2, Ganapathy Mani2, Kevin Kochpatcharin2, Pelin Angin2, James MacDonald4. 2020. Surveillance Video Querying With A Human-in-the-Loop. InWorkshop on Human-In-the-Loop Data Analytics (HILDA’20), June 14–19, 2020, Portland, OR, USA. ACM, New York, NY, USA, 6 pages. https://doi.org/10. 1145/3398730.3399192