Wednesday, October 5, 2022

Computer Vision AI: Parsing Concepts From Smarter AI and Qualcomm

I met with Chris Piche, CEO of Smarter AI, recently, and I’ve been following Qualcomm’s Connected Intelligent Edge effort for some time. 

In some ways, the two companies are potentially synergistic in that Qualcomm created and drives the hardware side of smarter edge devices, like cameras, and Smarter AI is focused on making those cameras more capable.  

Computer vision tools struggle with facial-recognition issues, bad training sets, and a tendency to misidentify people. But computer vision isn’t used only to identify people. It can be used to identify other things, like a person carrying a gun or a child doing something dangerous. 

Let’s talk about artificial intelligence (AI) as it applies to computer vision, Smarter AI, and Qualcomm.

See more: The Top Artificial Intelligence (AI) Companies

Qualcomm’s Connected Intelligent Edge

The big issues this effort can address are the need to reduce latency and the amount of data being pumped to the cloud for analysis. 

The first can make alerts overly slow and the second can result in larger problems as network congestion caused by these massive video streams significantly degrades site performance.  

By placing much of the intelligence into the edge device, it can not only potentially make decisions and send out alerts closer to whatever the camera is watching, but the transfer doesn’t degrade the overall network as much, resulting in a far lower drag on network speed and performance.  

The overall effort could improve our ability to observe the world in real-time and more quickly surface the benefits of any related security camera deployment. 

Smarter AI for potholes

Smarter AI is focused on computer vision. The examples they shared were of automated pothole scanners that could identify and report bad road potholes in real-time. 

The same capability could be applied to other objects, like looking for roadside bombs for the military, looking for defects that need to be fixed in and around buildings, and monitoring conveyer belts to identify and eliminate parts or assembly defects quickly and accurately.  

The technology could be used to minimize distracted driving, identifying and alerting when a driver picks up and looks at their smartphone, looking for unsafe roads that need help, beyond just potholes, and automated camera systems of city, county, state, and federal vehicles that are constantly looking for problems or behaviors and reporting them, without distracting the driver and allowing for those problems to be corrected, where, otherwise, they might have caused an accident.  

See more: How Artificial Intelligence (AI) is Used by 20th Century Studios, Epiq, PureTech Global, FintechOS, and WildTrack: Case Studies

Computer vision for law enforcement

The combined technologies from both companies could be critical in the creation of relatively crime-free smart cities of the future by identifying crimes in progress, people who are armed in places where weapons are not allowed, and an enhanced ability to identify crimes, such as from kidnapping attempts, so those threats can be addressed.

A safer future

While facial recognition has raised a considerable number of privacy concerns, using computer vision to identify objects could have a tremendously positive short-term impact.

It sidesteps the privacy issues, because it’s looking for things that don’t belong or are hazardous, instead of looking to break the latest untold story using an illicitly captured video. 

Computer vision and the intelligent connected edge promise a safer future, one where we can better understand and prevent unsafe behavior and the potential for cities to become smarter and more able to address their maintenance issues quickly.  

In short, both Smarter AI and Qualcomm could support a better, safer, and more data-driven world, which would have a positive impact on the quality of our businesses and personal lives.  

See more: The Artificial Intelligence (AI) Market

Similar articles

Latest Articles