There are many types of artificial intelligence, but one form of AI that has been quietly making waves in the background is computer vision (CV).

Computer vision analyzes images and videos and extracts useful data depending on the needs of the user. Or, to put it another way, CV examines visual data with the analytical approach of humans, but at computer speed. But there are some unexpected ways that computer vision is used, and some you've probably used without even realizing.

4 Unexpected Ways We're Using Computer Vision

Computer vision uses machine learning to analyze massive amounts of visual data rapidly. Many of us already use CV daily without giving it much thought. Did you know that you're using computer vision if you search through your photos for pictures of a dog or a beach or if you unlock your phone with facial recognition?

This is the public face of computer vision. But its use is becoming more widespread, and some of these uses may surprise you.

1. Content Moderation

Photo of polaroids depicting social media icons

Content moderation is a prickly subject crammed full of grey areas. While text moderation is a relatively simple concept that AI has been helping to moderate for years, the moderation of video and images still requires a more significant level of human input.

Now, some people may think that scrolling through endless social media posts seems like a perfect job. But the truth is quite shocking; these aren't pictures of puppies and somebody's anniversary dinner. AI can already quickly verify that these images are safe.

What this means is that the type of content that makes it through to moderators includes content that nobody in their right mind would ever want to see. There are numerous reports of moderators who have PTSD. A Harvard University article confirmed that moderators face considerable psychological risks.

Currently, the role of CV in content moderation cannot completely remove the human element. But with social media platforms finding moderation a virtually impossible task, CV can ease the burden. Computer vision is already used to greatly reduce the number of video "nasties" that filter through to human moderators. And, importantly, it can do so in almost real-time, reducing the risk of unsavory content reaching the eyes of the unsuspecting public and, hopefully, moderators.

2. Phishing Detection

Phishing attacks are potentially devastating for both individuals and organizations. Unfortunately, the process of keeping your systems and data safe against phishing attacks is an ongoing arms race between security professionals and the bad actors behind the attacks.

One of the problems facing security systems is a reliance on blacklists to identify the source of attacks. This is a reactive strategy. The problem with reactive strategies is the time lag between threat identification and the appropriate action being taken. This gap is what bad actors hope to exploit and is the same gap computer vision is filling.

CV is starting to be used as a real-time defense against phishing attacks. Instead of using blacklists to identify potential attacks, CV uses visual signals to identify possible red flags.

Some of the methods used to achieve this are listed below:

  • Identify spoofed websites
  • Identify trigger words disguised as graphics
  • Keyword padding and other text obfuscation

Although traditional security systems will remain on the frontline for the foreseeable future, the role of CV in plugging these shortcomings will be increasingly prevalent.

3. Monitoring Sports Sponsorship

Picture of a sports stadium

This one might seem like a curveball, so let's explain why this is important.

Sports sponsorship is huge, with billions of dollars spent yearly on sponsoring teams, events, and stadiums. One of the reasons that so much is spent is that sports sponsorship guarantees a captivated audience for the duration of an event.

In a world where advertisers often only have your attention for a few seconds as you scroll through your Instagram feed, a captured audience is like gold dust to marketers. The problem comes when trying to measure the effectiveness of a campaign.

Unlike digital campaigns, where performance can be measured precisely in almost real-time, the success of sports sponsorship is measured in a far more analog manner. With billions of dollars at stake, marketers understandably want more information about just what their money is getting them.

This is where computer vision steps in. For example, a company that advertises a racing car would use humans to monitor a race and count the screen time their advert achieved. This was laborious, time-consuming, and expensive. But now, many companies use CV to perform this task.

Additionally, it can be used to monitor the long-term success of a campaign. For instance, it can be used to determine how many times a video clip bearing their logo has been shared on social media platforms.

4. Counterfeit Detection

Picture of a credit card being passed into a computer screen

The internet is awash with counterfeit products. Many of these are sold by third-party vendors on otherwise reputable platforms. These platforms have legal obligations to ensure that the quality and pedigree of all products on their platform are as they should be.

For example, in 2020, Amazon destroyed over two million counterfeit products.

Monitoring counterfeit products successfully have always been problematic. Once again, one of the major problems is time. The lag between a product being listed and identified as a fraud can be long enough for the perpetrator to have shipped hundreds of products, taken the money, and disappeared.

This is the vulnerability that CV is being used to plug. It allows for real-time analysis of products listed on a platform's website. In addition, it analyzes various visual components to identify potentially counterfeited products. These include:

  • Logo Detection: This can identify products with illicitly used logos (sunglasses stamped with the Ferrari logo selling for a few bucks on Amazon, for example). Or poor-quality logos that give away the fact that those bargain Nike trainers might not be what they appear to be.
  • Image Analysis: CV can be trained to look for potential red flags like differences in color or labeling that may suggest a product is counterfeit.
  • Object Recognition: CV techniques can also recognize objects and patterns within images or videos. This can help to identify counterfeit products that have been altered or modified in some way, such as by changing the branding or labeling.

The counterfeit market is enormous and affects everyone, from the manufacturer to the end user. Using computer vision to identify counterfeits won't eliminate the problem, but it represents a big step in the right direction.

Seeing the Future Clearly

Computer vision is a rapidly evolving technology that promises much. Driven by factors such as the race to develop the first truly self-driving electric vehicles, the pace of development is relentless.

It is an exciting technology that will continue to throw out new and surprising uses as it matures.