Technology and Information Sharing

Automated Analysis of BWC Video to Evaluate Police-Community Interactions and Improve Public Trust

Location: Ballroom D1/D2 (Elm/Clear Fork), KBHCC

Jonathan M. Wender, Ph.D.

President & CEO
Polis Solutions
Dallas, Texas, United States
AM

Albert Martinez, M.A.

Executive Assistant Chief of Police
Dallas Police Department
Dallas, Texas, United States
Vivian Ta, Ph.D.

Chief Data Scientist
Polis Solutions, Inc.
Dallas, Texas, United States
Peter Tu, Ph.D.

Chief Scientist
GE Research
Niskayuna, New York, United States
PS

Peter Selfridge

Despite the widespread adoption of body-worn cameras (BWCs), there is currently no efficient, scalable way for police agencies to analyze the massive amount of video data they are collecting. This session presents an NIJ-funded project in Dallas, Texas, that is piloting the use of state-of-the-art technology called “multi-modal analysis” to automatically evaluate BWC videos. Multi-modal analysis combines natural language processing and computer vision to identify and evaluate key features of officers’ and citizens’ speech and behavior recorded in BWC videos. By combining multi-modal analysis with evidence-based models of procedural justice and community policing, this project will result in new tools that give agencies powerful ways to evaluate officer performance and measure community trust.