AI based facial identification system developed by NIT Warangal students

AI based facial identification system developed by NIT Warangal students

Posted by M.Venkatesh on September 20th 2017

Artificial Intelligence-based facial identification system which identifies people through disguises developed by NIT Warangal students.

A research paper titled ‘Disguised face identification (DFI) with Facial KeyPoints using Spatial Fusion Convolutional Network  submitted by two students of National Institute of Technology (NIT) Warangal – has been accepted for presentation at the ‘International Conference on Computer Vision Workshop (ICCVW) – 2017’ to be held on October 28th 2017 at Venice, Italy. 
 
According to a report published in The Hindu, G. Meghana Reddy and Devendra Patil – final year B. Tech students said that they had developed an artificial neural network with which it was possible to identify criminals sporting scarf, glasses, mask, beard, cap or a combination of all these. 
 
How it works
 
The deep learning-based system works by identifying 14 key areas of the face, including 10 for the eyes, three for the lips, and one for the nose. It is capable of estimating these even when they are obscured in some way. It then compares these readings to images to find a match. In early tests, the results were 56 percent accurate at finding the right person when their face was covered with a hat or scarf — although this dropped to 43 percent when they also wear glasses. Those figures are not going to be considered evidence in a court of law anytime soon, but they could certainly help police narrow down a search.
 
Termed as being a pioneering work towards crime detection their  paper has been accepted for presentation at the ‘International conference on computer vision workshop (ICCVW) – 2017’ being held on October 28 at Venice, Italy. 
These two students with their mentors – Amarjot Singh, an NIT Warangal alumni, who is currently doing his Ph.D. at Cambridge and S. N. Omkar, research scientist at IISc, Bangalore collected 10 disguises, 25 subjects, eight backgrounds and 2,000 images for their research. “This, in my opinion, is the first AI-based work that solved the problem of DFI with a reasonable accuracy,” Amarjot Singh said. Next, he notes that the team is trying to get the technology to function in real time with less computational power. “After that, the next step would be to deploy it on cameras to see how well it performs,” he continued.
 
Ms. Meghana said that this was first of its kind application in the world they chanced upon and they need to have more data to achieve greater accuracy. The police and other agencies would be having great data bank of criminals which will help to train the network. When a new image is fed to the computer using this artificial neural network, it would help show the real face of the disguised person. “We have done major part of our research at Innovation Garage on NIT Warangal campus, & this network detects the original face within five seconds. We could achieve 70% of accuracy and are still working on the project,” she said.
 
The research, published on the preprint server of  “arXiv” and shared in an AI newsletter, went viral after prominent academic and sociologist Zeynep Tufekci shared it on Twitter. 

Comments

No Comments Found

Leave your comment

Sign in to Post your Comments on this story