Heard industry leaders talking about
intelligent video analytics? It is revolutionizing the way public organizations
ingest and glean value from video content sources. Below, we break it all down
for you—from origin to application.
How it developed
The first intelligent video analytics
programs were primitive compared to today’s versions. Algorithms were applied
to video to detect motion and the location of that motion on live feeds. As
Frank Yeh, Senior Solution Architect for intelligent video analytics at IBM
explains in his blog, “Turning video into insight”, these early programs
weren’t actionable. Operators didn’t want or need an alert any time any motion occurred. There was no way to
distinguish between the motion that required notice and usual activity. “False
alerts” became all too common.
In a second, more advanced wave of intelligent
video analytics software, this was corrected. Modern programs provide great
detail on detected motion including color, shape, size, type and more. Frank
explains, “Instead of ‘something’s moving’ you now get ‘a red car is moving
eastbound on 33rd Street’ or ‘a bald man with eyeglasses wearing a red shirt is
walking down the hallway.’”
As more video cameras are purchased, placed
and used for public safety, the development and application of intelligent
video analytics has increased tenfold, enhancing a diverse number of
applications.
Why it works
How does intelligent video analytics
provide value beyond traditional surveillance systems? Curt Brobst, an intelligent video analytics evangelist at IBM,
explains that video monitoring systems relying on just human operators
to identify action of interest are inherently flawed. While recordings of video
can provide insight into an incident afterwards, they do not successfully
enable early detection and response while the incident is still occurring.
This is because we humans have limited
attention spans. The typical human attention span while watching video is a
cool 22 minutes. Even the most dedicated operator is still subject to “perceptual
blindness”, or the brain’s tendency to screen out actions in order to help
us focus. Even if an operator maintains focus, perceptual blindness could
prevent them from noticing unexpected activity.
As Curt explains, “Video analytics is never
sleepy, inattentive or distracted. It isn’t overwhelmed by trying to keep track
of dozens of video feeds. It isn’t affected by perceptual blindness. Rather, it
monitors all video feeds 24/7, notifying human operators when something of
interest happens.”
How it works
There are four key steps to developing and
achieving value from an intelligent video analytics surveillance
system—capture, ingest and analyze, decide, and act. All four are outlined in detail in this blog post. Implementing a
system according to these steps affords a number of benefits. Resources are
more effectively managed; as video sources increase exponentially, the cost of
personnel to monitor all incoming content would be prohibitive. An intelligent
video analytics system would manage and tag the incoming data, and make it
available and easily used by multiple departments—breaking down informational
silos. This allows departments to analyze and understand events before, during
and after they occur.
A few applications…
New applications and case studies are
appearing regularly as the value in this new information is really brought to
light. A few examples include managing traffic flow for a city—on a regular
basis and during planned or unplanned events like parades or emergencies. In public
safety, intelligent video analytics is helping respond to crime,
understand crime patterns and even to predict where and when resources need to
be deployed to have an impact. As Frank notes, with intelligent video analytics,
“retailers can better understand customer behavior, and banks and airports can
understand queue waiting times. The possibilities are virtually endless.”
Interested in learning more about how
analytics are creating a safer planet? Check out posts by experts on the IBM Big Data Hub.
No comments:
Post a Comment