Game over? Machine learning could help identify terrorist victory signs
Latest News »
- Govt to set up ministers’ panel to oversee PSU bank consolidation: Arun Jaitley
- Gold prices extend losses, shed Rs100 on muted demand
- Apex Frozen Foods IPO subscribed 1.2 times till 2.45pm on Day 2
- Suresh Prabhu takes moral responsibility for train accidents, offers to quit
- Stop dumping of ‘ritualistic material’ into Ganga to keep river clean: NEERI
It is often said that one man’s terrorist is another man’s freedom fighter. The political and moral dimensions of the debate aside, researchers have found a new method of identifying terrorists who cover themselves from head to toe.
Using machine learning biometric identification methods, Ahmad Hassanat and his team of researchers at the Mu’tah University in Jordan have worked out how to identify people or at least groups of people from the unique way they make the “V for victory” signs.
“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,” Hassanat told the MIT Technology Review. “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.
For the research, the team logged V-signs made by 50 men and women, of varying ages, combining into database of 500 images. These were photographed several times against a black background, using an 8-megapixel smartphone camera. They then analysed 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, which allowed them to identify various triangle shapes between these points, their relative size and the angles.
However, there are certain limitations here as well. The first is that this is a relatively small data-set at present, and this will have to be expanded on a much larger scale for it to be usable in the field and in conflicts. The second is that there is every likelihood of false positives and of misidentification of individuals until the algorithm becomes even more robust. Also, the system might get thrown off if the person they are trying to identify changes what they are wearing, such as gloves, or if there is a change in their weight.
Some other improvements need to be made. Hassanat and his team want to include information such as finger width and length to their machine learning system.
While this is not a perfect approach, it is possible that anti-terrorist agencies could use these image recognition systems to confirm the presence of specific terrorist groups, either in propaganda videos or even in certain geographical regions.