Skip to content

July 12, 2017

AI trained on 1990s blockbuster movies learns to identify handguns on video

by John_A

Why it matters to you

This artificial intelligence may be used to keep people safe by spotting handguns in places they shouldn’t be.

Humans are hardwired to quickly spot objects in the corner of our eyes: Kids getting into trouble, snakes slithering into the room, and the like. That task isn’t as easy for algorithms, but they’re getting better. Someday soon they may even help humans spot dangerous objects we might have missed. Researchers at the University of Granada, for example, have used deep learning to train an algorithm to identify handguns from video in real time.

The system may be used in home security systems, police body cameras, and video surveillance at places like schools, airports, and malls. It could also be applied to videos online as a way to monitor and map violent content.

Although around 30 percent of American adults say they own guns; the weapons have a steady presence in our lives. They seem to be everywhere you look — carried by cops on TV or as a key component in many of today’s most popular video games. So it’s appropriate that the University of Granada researchers trained their system on movies like Pulp Fiction, and the Mission Impossible and James Bond franchises, through which many of us also had our first introduction to firearms.

After training the algorithm on low-quality video of these classic films, it was able to effectively identify drawn handguns more than 96.5 percent, while analyzing in real time at 5 frames per second. When the algorithm spots a handgun, it borders it with a red box. The researchers think that the model can be combined with an alarm system in order to quickly and inexpensively identify firearms in places where they’re forbidden or may pose a threat.

Although the scientists don’t expect their model to completely replace current weapon-detection systems like metal detectors, they see it as a complementary system to help address some of the drawbacks of conventional methods. For example, metal detectors require guns to pass through a certain point, which — as anyone whose passed a security checkpoint knows — can create a bottleneck of people in crowded areas. The model is also able to discriminate between guns and other metal objects, which the metal detector cannot.




Advertisements
Read more from News

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Note: HTML is allowed. Your email address will never be published.

Subscribe to comments

%d bloggers like this: