Researchers from Carnegie Mellon University have created a system that can use video from smartphones to locate a shooter.
The system, known as Video Event Reconstruction and Analysis (VERA), requires as few as three videos to accurately locate a shooter, as demonstrated using three such recordings from the 2017 Las Vegas shooting that killed 58 people. In this case, the system was able to accurately trace the shooter's location to the north wing of the Mandalay Bay hotel. This result was determined using only three gunshots from the first minute of the shooting.
Alexander Hauptmann, a professor in CMU's Language Technologies Institute, was careful to specify that the system won't necessarily replace the dedicated microphone arrays that police currently employ, but of course, these arrays aren't always as readily available, and thus, the system is seen as a welcome supplemental tool. Another possible use is for journalists and human rights workers.
The system uses machine learning technology to synchronize multiple video feeds, then to calculate positions of each camera, then uses the audio from the feeds to find the time delay between the bullet's shock wave and the muzzle blast. These differing time delays can then be used to find the shooter's location. The audio information can even be used to determine the type of gun used by the shooter and thus, the speed of the bullets.
The research team has released the system as open source code. You can read more at Carnegie Mellon's website.