Drones Could Be Used to Spot Violent Behavior in Crowds

June 13, 2018 11:16 pm Last Updated: June 13, 2018 11:16 pm

Drones are being used as automated surveillance to detect crime and violent behavior. Researchers are developing ways to use machine learning to analyze violent behavior in live footage with camera-equipped drones.

In a research paper from the Department of Engineering at University of Cambridge, titled “Eye in the Sky,” it was revealed that the system uses a Parrot AR quadcopter to transmit live footage through a mobile internet connection for analysis. An algorithm uses deep learning to identify poses that are designated as violent, which includes strangling, punching, kicking, shooting, and stabbing.

Amarjot Singh, the lead researcher from the University of Cambridge hopes that this new surveillance method could possibly prevent attacks like the Manchester Arena bombing in 2017.

However, there are still questions regarding its accuracy as a nonviolent gesture might be misinterpreted as violent. Other ethical concerns include how people will constantly feel that they are being monitored and analyzed.

Despite these criticisms, Singh claims that it could have a positive impact on stopping terrorism and that anything can be used for good or bad.