Do you dream to live in the city where you can walk late at night in the local park fearlessly?
Do you want to stop being afraid of the child coming home from school?
Just admit it: you’ve been dreaming to take your grandson to watch football game, and everything that confuses you are the fan fights in the stadiums and nearby. You don’t want your boy watch this.
What about garbage in the streets? Aren’t you tired of it? Someone throws it!
Someone breaks the newly-planted trees, someone makes the buildings dirty inside with obscene writings. Do you want to stop it?
Everybody wants. We won’t exaggerate demanding that everybody wants to live in the clean and calm city or small town without hooligans drinking alcohol while hanging about the street, without garbage on the roadside and broken public property.
Of cause, there are uniformed guys patrolling the streets. But they cannot check all the possible areas in the town, moreover only few of them are in charge and it’s hard to hire more and more policemen to put them on every corner after each 100 meters.
The government can use cameras, certainly, but what next: the police will check the camera ones, twice will send the policemen to detain some hooligans or law breakers… But right the next day the situation will repeat again, the police will come and detain again, but nothing will change seriously, and your personal dream of the finally calm and nice hometown will never come true.
However, there is a solution. And, as it is usual today, this solution is associated with high technologies. At this time at the service of law, order and security stands face recognition technology.
Just do not rush to brush off, thinking it was something from science fiction, that will be put into practice in 50 years, and then only in major metropolitan areas around the world. The technology itself is quite really exists, and the future is now.
Here comes the solution
Russian NtechLab company with its best specialists in the field of machine learning has developed one of the most accurate and fast facial recognition algorithms in the world, available in the form of SDK, which can be built on the client side and in the form of Cloud API, that handles requests on the side of the developer.
That’s true — the algorithm is indeed one of the most accurate and fast at the moment in the world. One example of this — Washington MegaFace contest-2015, during which NtechLab algorithm showed the best result, 90+ beating contestants from all around the world, including Google team program.
To understand how the algorithm of the future works and how it relates the solution of the security problem, you need to briefly look into the past and turn to the most basic things — technology, its emergence, finally understand where it came from and what it stands for.
Neuron networks for newbies
Face recognition technology is based on the artificial neural network which in its turn based on the principle of a biological neural network.
Neuron is a small brain cell. The essence of the neuron is in the accumulation and transmission of information to other neurons. Neurons are connected with each other via axons and dendrites, forming some sort of coil. This tangle of intertwined with each other small sources of information is called neural network. Information generated and transmitted by neurons can be very different; each area, or fragment, of the brain is responsible for a certain function of a body, i.e. motor or visual. In case of facial recognition we are interested in the latter.
Scientists applied to the biological neural networks research at the end of the 19th century. But only after dozens of years there is appeared interest to the artificial neural network modelling. Why did people need it?
Probably, that’s people’s nature — we always need to create something more efficient, something perfect. And if it is life with all its errors and imperfectness, people need to create its more efficient likeness, more workable prototype of it. That’s how science fiction writers and filmmakers came up with their futuristic objects, which are now becoming real — emotional robotic nurses, drones, virtual reality etcetera.
The creation of artificial intelligence, which will never get tired, make a mistake, on the whole, emerged from the same need to create a similarity of life, which in many ways will be more perfect then the «original pattern».
One of the key features of the artificial neural network is its ability to learn (deep learning). The artificial neural network training is somehow similar to a biological neural network training. Both learn by making mistakes. By trial and error in human behavior are laid certain rules and models.
The artificial brain is constructed by specialist, then it is checked to be correlated, the algorithm becomes more precise, the errors are gone and — Viola! — The algorithm is set to perform a specific function.
Remember, we talked about the network fragments-areas, which are responsible for different functions? The same happens with the artificial neural network. The only difference is that each fragment of the artificial neural network is responsible for only one function (for example, the ability to see an object and compare it with many others). Also it is not located in the brain but is programmed as part of a task, for example, the task to compare facial images. For other tasks, such as language recognition, the new algorithm will be programmed.
And now you’ll probably ask: So what? Where is the connection with security matter, cleanliness, and calmness? Aren’t we too far from the stated topic? Not at all!
The fact is that the algorithm created by the NtechLab team is trained for the face detection so accurately that any surveillance camera, the video stream from which is processed by the NtechLab technology of facial recognition, suddenly turns into a powerful tool for tracking and apprehending all the above mentioned hooligans and troublemakers .
What actually NtechLab offers, and how it helps with security matter?
For now its FindFace.pro product line is represented by FindFace Cloud API и FindFace Enterprise Server SDK.
FindFace Cloud API is a cloud face recognition API for integration into third-party solutions with all files stored on the secure FindFace Cloud while FindFace Enterprise Server SDK is an on-premise face recognition API.
There are too many wanted individuals in the average city for the police to remember all of them. It’s impossible to check everyone’s ID.
The Ntechlab cloud technologies allow the solution:
Face detection and identification in a video stream
Including surveillance CCTV cameras in public spaces
Restricted site access and database search
E.g. for fans blacklisted from a stadium
Personal identity verification
Including entrance facilities and mobile police officers
Monitoring movement of people
Throughout the city or facility
The competitive advantages of NtechLab product line are:
The world’s one of the best accuracy
- 95% rank-10 probability among 10K photos
- 88% rank-10 probability among 1M photos
- 99% verification accuracy
World’s best performance
- Half-second search time for 1 billion photos
- Unique index performs very fast search while maintaining the highest accuracy
- The index needs just 20 GB for 500 million photos
The city can be safer thanks to the automatic identification of suspects and potentially dangerous people using the algorithm, processing cameras video stream!
No matter, whether the surveillance cameras integrated into the urban landscape, or transport facilities, into critical infrastructure, or government sites. Face recognition technology automatically identifies all the known suspects and alerts authorities.
The law breaker will only have to wait the doorbell or the penalty info on special website or in the mailbox.
However, the security problem is not limited to law breaking.
Restricting access to objects for those in «black» lists, the identification of potentially dangerous people, work with patterns of behavior and emotions… yeah! the scenario of sci-fi blockbuster is about to break out on the streets of your hometown!
Photo or video are often the only evidence of the crime. Regular search database requires a lot of time and attention, the errors are almost inevitable.
The algorithm by NtechLab team is able to search among millions of photos in less than a second. You can identify each person in the photo or video, even if since the last shooting he grew a moustache or a beard, even if photo was shoot in low light and long time ago.
Do not believe us?
Come on visit findface.ru, upload photos in a special Web interface and believe it yourself that the future is now.
Read us more about the most incredible areas of applications of the cloud face recognition technology by NtechLab. New stories will soon be here. Stay tuned!