The National Crime Records Bureau (NCRB) has launched a revised tender for a nationwide automated facial-recognition system. The revised document states that, “this project does not involve installation of CCTV cameras nor will it connect to any existing CCTV camera anywhere”.
This is a significant change from the earlier model of the request for proposal (RFP), which closely emphasised on extracting information from CCTV cameras. A practical requirement within the earlier RFP — “capture face images from CCTV feed and generate alerts if a blacklist match is found”, — has made approach for the much less particular requirement of “interfacing ability to analyse data on images/visuals on crime and criminals from various sources”.
The final date for submission of bids in accordance with the recent RFP is August 6, 2020. That makes it the ninth revision in dates for the reason that earlier RFP got here out mid-last yr. A pre-bid convention is scheduled for all of the potential bidders on July 13, 2020.
Currently there isn’t any authorized framework in place for the implementation and enforcement of such applied sciences and the revised doc doesn’t point out if there may be any dialogue on the anvil concerning the legality of such programs.
While the brand new RFP doc states the usage of photographs and movies in bodily and digital type to hold out searches, the point out of curating a database via photographs from newspapers, raids, and many others. has been scrapped.
The velocity and accuracy parameter of the facial-recognition system has been revised to incorporate fewer take a look at instances as compared with the earlier model. For a velocity take a look at on non-ingested video to nonetheless picture, the earlier model had 50 take a look at instances for movies of 300mb and 10-60 seconds every to be examined in opposition to the velocity of 10 seconds, however in accordance with the revised model, the velocity take a look at will now be performed on solely 10 take a look at instances of 10-300 seconds every on the fee of 10 seconds for a 1 million NIST-format database. Similarly, the velocity take a look at case has been decreased to 10 take a look at instances from 50 (nonetheless picture to non-ingested video) and 30 (video to video) take a look at instances in opposition to 180 seconds.
The revised model has additionally eased the pre-qualification standards to some extent. It states that the bidder have to be registered underneath the Companies Act 1956 or 2013 and have an annual turnover of INR50 crore in every of the final three monetary years, not like INR100 crore beforehand. However, startups recognised by the Department for Promotion of Industry and Internal Trade have been exempted from this clause.
Any worldwide participant or a companion of a joint-venture firm keen to take part within the bidding course of should present proof of handle relating to its presence in India. The bidder ought to have appeared within the NIST Facial Recognition Vendor Test analysis programme performed by NIST 2016 or later, with at the very least three automated facial-recognition system installations in legislation enforcement or authorities businesses worldwide within the final 10 years.
This qualifies main worldwide face-recognition software program suppliers just like the Japan headquartered NEC Corporation. It stays to be seen how Indian startups, with comparatively smaller databases of what’s a reasonably new expertise, will handle to compete.