New A.I. Device Will Activate Cameras and Alert Police When It Detects Crime-Related Noises (Gunshots, Glass Breaking, Tires Screeching)


by BN Frank, Activist Post:

Experts have warned for years about using Artificial Intelligence (A.I.) technology (see 123).  Embarrassing as well as tragic examples of A.I. inaccuracies continue to be reported (see 1234).

People have been accused and convicted of crimes based on inaccuracies (see 12) including from the use of A.I. based ShotSpotter technology.  Nevertheless, a new A.I. device is being marketed to American communities and police departments.

From GovTech:


Exclusive: Flock Safety Reveals Gunshot Sensors Tied to Cameras

By Ben Miller

The device, called the Raven, is going through beta testing now and will launch for general sales in January. It’s designed to detect gunshots, as well as other sounds such as glass breaking, and activate nearby cameras.

A gunshot rings out, and overhead a camera begins recording.

With a new product, this is what Flock Safety hopes to achieve. And more than that, it wants to make the product cheap enough and fast enough to set up that any local police department can use it.

The product — going through beta testing now and hitting the market for general sale in January — is an audio sensor called Raven. It’s meant to detect gunshots, along with other crime-related noises such as glass breaking and tires screeching, and then activate nearby cameras as well as alert the police.

Garrett Langley, co-founder and CEO of Flock Safety, hopes Raven will allow law enforcement to respond to crimes faster and relay more information to detectives. In a best-case scenario, Raven could put police on the scene before an incident escalates.

The camera tie-in is critical. Gunshot detection is nothing new — ShotSpotter has been doing it for years — but the whole selling point of Raven is that it will connect to Falcon, Flock Safety’s camera that helps police locate vehicles of interest.

“The audio event is critical. But to me, it’s that integrated solution about how are we actually going to solve homicide, how are we going to prevent a homicide?” Langley said. “And that’s what we’re selling. So could it work alone? Sure, but we would never sell it alone, because I think that that would be a truly ineffective solution.”

Like the Falcon camera, the Raven audio device is solar powered, so it doesn’t need to tap into an existing power source. Raven can trigger cameras to begin recording, but it relies on cameras to do the actual processing to determine what kind of sound has occurred. Unlike ShotSpotter, which has a center staffed with people working to validate that a gunshot was, in fact, a gunshot, Flock Safety plans to rely solely on its AI algorithms to identify a sound.

“Something loud or something with a particular frequency happens in a particular area … then we start recording, and then we start filtering through the machine learning on the edge to decide, in more or less granular buckets, this is a gunshot, this is glass breaking, [these are] some of the things that we’re interested in. And then there’s everything else, which is all the things we’re not interested in, [like] human voice and all this other stuff, and that stuff is just dumped immediately,” said Davis Lukens, Flock’s vice president of product. “Then once we have it [through that] first set of filtering, that’s when it gets backhauled to the Falcon, and then sent off to the cloud for additional processing and more granular detection of — you know, maybe it is a firework that sounds very similar to a gunshot.”

Lukens declined to give an exact accuracy rate — which will change between different environments, and likely between different types of sound — but said an aggregate accuracy rate would be north of 90 percent.

Even though the tie-in with cameras means a possible extra layer of validation for whatever sounds the Raven picks up, the very nature of the product is at odds with the criticisms racial justice activists have been making for years. Oftentimes cities will deploy technologies such as license plate-reading cameras and gunshot sensors in areas with more low-income residents or people of color, which means that those technologies will tend to send police to those areas more often. As a result, the department will log more crimes in those areas, which will make it seem like those areas need even more policing.

Read More @

Source link

You might also like
Leave A Reply

Your email address will not be published.