Face Detection Engine

Face Detection is about finding faces in arbitrary scenes in images and videos. Very much research is going on in this area, and the number of applications (facial recognition, marketing, photography, social networks, public security) has rapidly expanded. Almost every week we get a message from people all over the world, asking us more or less the same question: I’ve got a photo of a person, can you help me find him/her on the Internet? The first step in automatic facial recognition is the accurate detection of human faces.

Face Kuznech’s Face Detection Engine determines the locations and sizes of faces on your digital images and videos. Working off of previous face detection approaches in addition to the company’s own comprehensive research, Kuznech produced one of the most effective and accurate face detection algorithms ever seen: the technology is smart enough to detect faces even at different orientations.

Try demo

Benefits of Kuznech Face Detection for you:

  1. Enjoys the highest precision* and lowest false alarm rates (percentage of detected faces that are non-face)
  2. Supports multiview face detection (when faces are rotated or captured under challenging conditions)
  3. Is effective in poorly lit conditions
  4. Works with cluttered backgrounds
  5. Localizes and detects faces even through glasses, facial hair, different hair styles, and heavy makeup

* As compared to solutions currently on the market.

Want to learn more? Read our Frequently Asked Questions or contact us.


Face detection is a special computer technology, that is used to determine the existence, location and size of human faces in digital images and videos. For still images, it detects the facial features and “isolates” the face from the rest of the picture, and – for videos – tracks a given face (person) in the flow of video frames.

We, humans, can process faces very quickly, usually within a split of a second. For computers it is a somewhat more complex and challenging process, because human faces are non-rigid objects with a high degree of variability in size, shape, color and texture. Attempting to detect a face, a computer will start by checking whether it is a still photo or a video image. Then, it determines if there are any faces present in that picture and distinguishes these faces from any other objects (patterns) in the background: bodies, trees, animals, furniture etc. This task has to be completed regardless of positions, scales, orientation, lighting conditions or camera distance.

A computer can detect faces task with the help of following methods:

Face Detection in images with controlled background
In this method, you simply have a frontal face image against a plain mono color background. Removing the background with special software gives the face “natural” boundaries. This is the easiest possible method.
Face Detection by color
Detection of skin color in images is a very popular and relatively simplistic technique for face detection. It obviously requires that the photos or video images are color (and not monochrome). The software scans the picture looking for areas that are of a typical skin color, then looking for face segments. Limitations with this technique are that skin color varies from race to race, so this method does not work as well for all skin colors. Besides, the color of a face (the exact hue of a person’s skin) varies with changes of lighting conditions, which can impact the correctness of facial detection, too.
Face Detection by motion
Faces are usually moving in real-time videos, so movement can be used as a guide by capturing the moving area. Of course, other objects in the scene (not only faces) can also move, so the software needs to look for particular reference points to indicate that it is actually a face that is moving.
Face Detection using Viola-Jones method
2001, Paul Viola and Michael Jones, two computer scientists, proposed a revolutionary method of computer face detection: they developed such a fast (15 times quicker than any technique at the time of release!), simple and robust (achieving high detection rate) face detection algorithm that it was soon built into standard point and shoot cameras. Viola and Jones did not try to analyse the image directly: instead, they started to analyze rectangular “features” (also known as “Haar-like features”), due to the similarity of the analysis of complex waveforms with Haar wavelets) in the image. These are simple square waveforms, and are named after a Hungarian mathematician Alfréd Haar.
Face Detection using convolutional neural networks method
In real world, face detection often sets two conflicting challenges: an advanced model to accurately differentiate faces from the backgrounds and computational prohibitiveness of effective models for the problem. To overcome this obstacle, a cascade architecture built on convolutional neural networks (CNNs) is used: on the one hand it has a very powerful discriminative capability, on the other hand it maintains high performance. CNN cascade operates at multiple resolutions, quickly rejects the background regions in the fast low resolution stages, and carefully evaluates a small number of challenging candidates in the last high resolution stage.
Kuznech’s Face Detection system uses all of the above methods combined with the own comprehensive research to make face detection both accurate and efficient. Still have questions? Contact us!

You can detect faces on your own images in Kuznech’s Face Detection and Recognition demo.

Best for

Image stocks
& Photo sharing sites

Manage your image database by linking pictures to people.
Sounds interesting? Contact us to learn more about the Face Detection Engine’s features and pricing plan.

Face Recognition Engine

A person’s face is a rich source of information. Simply by looking on a human face it is possible to tell, whether a person is male or female, approximately how old they are. Not anymore such analysis is a uniquely human capability: computer vision can do this, too. Kuznech’s Face Recognition Engine is used to automatically identify a person in a digital image or video source and determine the person’s main characteristics, such as age and gender.

Try demo

Benefits of Kuznech Face Recognition for you:

  1. Accurately recognizes faces even when they are very small or are turned 30 degrees on a plane
  2. Recognizes such parameters as gender (male/female) and age group
  3. Not affected by differences in lighting or the facial expressions of the subject
  4. Allows for the creation of a large gallery of models and identifies people from that list
Still have questions? Read our FAQ or contact us.
Face recognition is often confused with face detection, although they are very different in reality. Face detection identifies an object as “a face” and locates it in the image or video. Face recognition, in its turn, has to decide if a “face” is someone known or unknown, establish its identity and distinguish features, e.g. person’s age and gender. For face recognition to occur, there must be face detection first. To link together pictorial personal data (e.g. a face) with textual data (e.g. name), face recognition software prerequisites a pre-existing database of faces.

Face recognition is an important task used in a wide variety of security related systems, such as building access, border controls with biometric passports, in video surveillance, or for authentication to computer systems.

You can recognize faces on your images in Kuznech’s Face Detection and Recognition demo.

How it works

  1. The engine localizes a face in an image with the help of Face Detection Engine.
  2. The engine detects facial features (peaks, valleys, and landmarks) and treats them as measurable nodes.
  3. The system determines the person’s characteristics (like gender & age).
  4. Facial recognition algorithms suggest who the person in the picture might be.

Best for


Understand the interests of your audience and help advertisers pinpoint target demographics.

Security and surveillance systems

Use Face Recognition algorithms to verify identities and avoid fraud.
Contact us to ask any questions you might have regarding our Face Recognition Engine.


Face Detection and Face Recognition technologies are licensed by size of image database and intensity of use (number of search queries). Please contact us to learn more about Kuznech Face Detection and Face Recognition pricing conditions.

Our other products:

  • LikeThis

    Recognizes products and finds the similar products in your catalog
  • Logotrakr

    Detects brands in a videostream. Enables to monitor brand citedness
  • ShelfMatch

    Automated control of window displays in brick-and-mortar shops
  • Adult Detection

    Pornographic scene classification, logo and skin detection