Skip to Content

Machine-Learning Algorithm Aims to Identify Terrorists Using the V Signs They Make

Terrorists often use masks, scarfs, and hoods to hide their identities. But a new approach aims to distinguish them using the shape of their fingers when they make the “V for victory” sign.

Every age has its iconic images. One of the more terrifying ones of the 21st century is the image of a man in desert or army fatigues making a “V for victory” sign with raised arm while standing over the decapitated body of a Western victim. In most of these images, the perpetrator’s face and head are covered with a scarf or hood to hide his identity.

That has forced military and law enforcement agencies to identify these individuals in other ways, such as with voice identification. This is not always easy or straightforward, so there is significant interest in finding new ways.

Today, Ahmad Hassanat at Mu'tah University in Jordan and a few pals say they have found just such a method. These guys say they have worked out how to distinguish people from the unique way they make V signs; finger size and the angle between the fingers is a useful biometric measure like a fingerprint.

The idea of using hand geometry as a biometric indicator is far from new. Many anatomists have recognized that hand shape varies widely between individuals and provides a way to identify them, if the details can be measured accurately.

However, the task of recognizing people using just a section of their hands is much harder. “Identifying a person using a small part of the hand is a challenging task, and has, to the best of our knowledge, never been investigated,” say Hassanat and co.

The team began by asking 50 men and women of varying ages to make a V sign with their right hand and photographing it several times against a black background with an eight-megapixel camera phone. This produced a database of 500 images.

An important question is how much information should be extracted from these images to aid in identification. Hassanat and co point out that many real-word images have low resolution, which limits the amount of information that can be gathered.

So they limited their analysis to determining the end points of the two fingers, the lowest point in the valley in between them, and two points in the palm of the hand.

This allowed them to analyze various triangle shapes between these points, their relative size and the angles they make, and so on.

They also used a second method to analyze the shape of the hand using a number of statistical measures. Combining these two methods creates a total of 16 different features that can be used in identification.

They then used two-thirds of the images to train a machine-learning algorithm to recognize different V signs and used the remaining images to test its efficacy.

The results make for interesting reading. Hassanat and co say that the combination of techniques allows them to distinguish the people in the data set with an accuracy of over 90 percent in some cases. “There is a great potential for this approach to be used for the purpose of identifying terrorists, if the victory sign were the only identifying evidence,” they say.

There are limitations to this work, of course. The first is that this is a relatively small data set, and Hassanat and co will want to show that their method works on a much larger scale. The second is that the likelihood of false positives and negatives has not been analyzed in detail. How likely is it that their algorithm misidentifies individuals?

That’s something the team is no doubt thinking about. And there are certainly improvements that can be made. Hassanat and co want to include other information in their analysis, such as finger width and length.

Of course, recognizing somebody by a V sign doesn’t give you a person’s identity. For that, the information would have to be combined with other data. Nevertheless, this is curious work that reveals how the pressures to identify nefarious individuals in the 21st century are leading to ever more inventive biometric techniques.

Ref: arxiv.org/abs/1602.08325 : Victory Sign Biometric for Terrorists Identification

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.