In some parts of the world, gunshots are heard more often than in others. Sometimes in places where one shouldn’t be hearing gunshots. Places where response teams need to act fast to catch the person firing them. A team of researchers at Carnegie Mellon University have developed a system that can use smartphone video to locate the source of gunshots using machine learning.
The system is known as Video Event Reconstruction and Analysis (VERA) and can work with as little as three videos to accurately locate the position of a shooter. As demonstrated in the video above of the 2017 shootings in Las Vegas that killed 58 people and wounded 413, it was able to accurately determine the shooter’s location to be the north wing of the Mandalay Bay hotel.
The system works by estimating the distance between the gunshot and the camera. It does this by using the time difference between the sound of the gunshot’s muzzle blast and the shockwave hitting the phone’s microphone. With multiple simultaneous recordings from different vantage points and AI learning technology to synchronise them together, they can triangulate where the source of the sound is.
The results from VERA aren’t as instant as those using elaborate microphone arrays often used by law enforcement or military to locate shooters quickly, and its accuracy will depend on how many phone cameras are shooting at the same time.
Right now, though, they say it’ll most likely be used for after the fact forensics, given its current speed and technical limitations. But in a rapidly unfolding situation in a populated area, this could be a valuable tool to help locate a shooter until more advanced technological help arrives.