The horrific 2012 Delhi gangrape case, which had sent shockwaves across the nation, had prompted scientists from the city-based Centre for Development of Advanced Computing (C-DAC) to start working on a next-generation safety application.
The app, which includes a voice-enabled alarm system, is called ‘Abhayam’ and it aims to provide emergency help to those in distress. Now, the team behind Abhayam is trying to come up with an app that will incorporate facial recognition as one of the safety features. As the number of crimes against women continues to rise, numerous mobile-based applications have been developed, but many of them need an activation button to start working.
“For any person in distress, there is very little time to react and activate a button on the safety app. The advantage of Abhayam is that it can be activated using a voice alert,” said Sandeep Kumar Srivastava, head of C-DAC’s emerging solutions and e-governance division.
With the availability of an in-house language expert team, the app development team has incorporated voice-enabled codes. When a person says ‘bachao’ (help), the app tracks down him/her and also sends SMSes to multiple mobile numbers, registered as emergency contacts on that phone. The app also has a provision to make automatic calls to the ’emergency contacts’, and it is enabled to auto-videograph and auto-photograph the scenes at the site from where the distress signal was sent.
“While testing the application, we asked the subject to shout ‘bachao’… the application immediately shared the location co-ordinates with both the registered ’emergency contacts’ and the police,” explained Srivastava. Though the application is ready for use, a final decision by the Ministry of Electronics and Information Technology (DeitY) is pending.
As part the other project, technology experts at C-DAC are “studying” videos to understand crowd behaviour, particularly during violent incidents or crimes. It is often noticed that social media sites are flooded with videos depicting attacks, mob rampages, or accidents, captured either by CCTVs or mobile phones of onlookers. But it is often difficult for investigating agencies to extract clues from the moving images.
“Using the new video surveillance application, shorter versions of the videos would be made available, making it easier for police to zero in on, and focus on the most vital portions,” said S S Kadam, one of the team members working on the Content-Based Image Retrieval System.
This team is also working on studying facial expressions and the behaviour of the crowd from these live videos. “It is possible to retrieve up to 128 facial expressions from a single image. It can be used to investigate violence or acts of vandalism at public places,” said Srivastava.