G’day! Long time no see and this means you deserve something special. And today our special guests are… Mr. Cloud and Mr. Server. Actually, in some situations, they really can help and save much: time, money, life.

All the time since we started our weblog we’ve been talking about neural networks, machine learning, facial recognition systems, metrics and evaluation methods. It’s now time to tell about our products. Moreover, recently the new version of our major product FindFace Enterprise Server SDK has come out. But first a little bit of history, ladies and gentlemen.

Technology becomes a product

When the NtechLab algorithm was created the authors wanted to test it on the image database of a massive volume.

At that moment the MegaFace one million photos database fitted good for testing our facial recognition software, and in September 2015 the NtechLab team sent its algorithm to the MegaFace Challenge.

According to the results, published in December 2015, the algorithm was deemed the best.

Up to that time, we knew that the technology can be and must be converted into products, and we knew how to do face recognition. But first, we needed to show the algorithm in real work, to demonstrate future users the real case, useful, interesting and admirable. We needed an audience.

Thus, jointly with our partners, we launched a service called FindFace.RU. Anybody regardless of his/her faith, race, level of education and political views could upload anyone’s photo to the website and find a similar person (the very person on the photo as a rule) in the Russian social network VKontakte.

The service worked great and in a short time became very popular. Potential clients and partners contacted us from all over the world. What is more, they suggested their own scenarios of future product applications: from social networks to retail and security systems to identify offenders and report them to the law enforcement authorities.

Cloud advantages and specialties

It was obvious that people needed to access face recognition services for their businesses. As a result of the FindFace.RU boom, we launched our first product — the system called FindFace Cloud API, which could be very useful for those potential clients and partners who wanted not only to test the algorithm but also to do something based on it by themselves.

The cloud platform gave the opportunity to integrate our technology into third-party solutions by REST API. Via API were sent requests for the key tasks such as verification and database faces search, and now these requests include age, gender and emotion recognition.

At the moment of launching the cloud product we certainly understood that it had its own restrictions: huge resource consumption when working with the video streams, the need to store the data in the cloud.

Moreover, it was obvious that different business tasks needed different solutions. If you don’t need to work with video and you have only photos, then FindFace Cloud API is just the thing: it is incredibly easy to work with, it does not need to deploy the server, you do not need to install anything. The cloud solution perfectly suits developers of web services, various mobile apps, etc.

However, market demand is not limited to such tasks. Many companies that wish to implement the facial recognition technology in their business solutions may need a product that will:

a) work within the customer’s infrastructure

b) work with video streams (the cloud opportunities, even if such an option will appear in the cloud, are limited)

Server advantages and specialties

Just after launching FindFace Cloud API the company released FindFace Enterprise Server SDK.

This principally new product could be easily integrated into the infrastructure of any company interested in the opportunities of their business using face recognition technology.

The deployment of FindFace Enterprise Server SDK on both an individual server as well as a distributed network of servers

The server product was developed for much more complicated client systems and is useful for big businesses that pay attention to privacy and work with video streams: banks, security, retail chains, etc.

Nevertheless, it is worth mentioning that the integration of FindFace Enterprise Server SDK will not demand from any software engineers some special skills in the field of neural networks or deep machine learning.

Today FindFace Enterprise Server SDK offers a wide range of capabilities, including face detection, verification, identification, and detection of age, gender, and emotions. It implements the cross-platform REST API, so any application can call it from the web, desktop, or mobile. Despite being a cloud product, it works in the client’s private infrastructure.

The new version of the server product supports video in FindFace User Interface (FFUI):

Video stream support in FindFace User Interface

The brand new video detector can now operate for the particular area of a frame, which makes the whole work with video streams less resource-intensive. It also allows for working offline via USB-dongle. In the case of cluster installations, all the server components now refer to a single local server.

You can ALSO appreciate:

  • The ability to make black and white lists, based on current databases
  • Grouping images with the same face into «person»: in order to find all the images of a certain person, joined together under unified Person ID
  • The new local server with a web interface allows you to control license according to its working period, the chosen parameters and functionality.

the National Institute of Standards and Technologies (NIST) of the U.S. Ministry of Commerce. According to the results, the NtechLab algorithm was acknowledged as the best in the world.

What is NIST, why the developers from all parts of the world send their algorithms there, how the testing is going on, and what the NIST certificate means if you decided to get involved in face recognition — we’ll tell about it in the next posts. Stay tuned and don’t forget to leave your comments below.