DailyPost 3039
FACIAL RECOGNITION – THE OMNIPOTENT POLICING
Video footage of the right quality and coverage emanates out of a CCTV network, which is meticulously technically planned and operated, as per the requirements. It is much more complex than it sounds, and video analytics are coming into play in a big way. There are endless moving parts in this system, which has to be first neatly understood, before its execution and operation. Field level continuous testing holds the key, which should continue till all the stakeholders are satisfied. We still have a long way to go, even to reach fully automated facial recognition capability on our CCTV systems, which has the potential to be one of its kind force multipliers.
The utility of video footage is being proven over and over again. It played a critical role in identifying and capturing Ajmal Kasab during the 26/11 Mumbai attacks. In the recent RG Kar Medical College & Hospital rape and murder case, it was through video footage Sanjoy Roy, the accused could be identified. In the most recent attack on Saif Ali Khan too the accused was identified through video footage of CCTV system inside the building and further clues to investigation came from facial recognition security cameras at Bandra railway station. Any number of cases have been detected through facial recognition in this manner. Suffice to say, even in the most crude and nascent form, video footage has proved its worth.
Think of a situation wherein a face could be detected in real time across all cameras in a city. It could happen across cities if they were to run similar connected CCTV systems enabled by state of art facial recognition software. Today the footage is being recorded and recorded and nothing is done with it, otherwise, but for traffic offences, it should have been used for any and every case where an accused has to be traced. There are issues of networks, tech and bandwidth today as well but are we moving in the direction of having an omnipotent and active monitoring of faces across the city/ies.
It should turn dynamic and real time if a face is to be detected in that area. There are any number of use cases in regular police work to monitor different nature of criminal elements; alerts and reports in this regard should become the order of the day. The nature of video analytics required has to keep on improving and so also the business logic. What about the system raising an alarm if any moving person with a covered face is detected? If facial recognition is used on the security guards themselves, it would have a transformative impact on the nature of services being delivered by them. What about artificial intelligence being used for this purpose, the multimodal LLMs can do wonders. Now you can interact / talk to a video. The possibilities to improve efficiency in policing by facial recognition are endless, we have to start thinking in the right direction.
PHYSICAL BODIES, FAST MOVING VEHICLES AND CRUDE TECHNOLOGY DOES NOT ADD UP TO PROFESSIONAL POLICING.
Sanjay Sahay