Microsoft Project Oxford API (Face API)

Face Detection:

The Face API provides high preciseness face location sightion that may detect up to sixty four human faces in a picture. Face detection may be done by uploading a complete JPEG file or by specifying a computer address of Associate in Nursing existing JPEG image on the online.
The detected faces area unit came with rectangles (left, top, dimension and height) indicating the placement of faces within the image in pixels.

It can be considered as a case of Object-class-Detection.
Face detection includes gender age etc can be done through state of Art algorithms, genetic algorithms, and the eigen-face Techniques.  These algorithms process face images and apply the operation of age prediction, gender, recognition, verification, similar face searching, grouping and face identification.

Face Recognition:

In general, face recognition provides the functionalities of mechanically distinguishing or validating an individual from a variety of detected faces. Through this application face is identified or verified from a digital image and video. It can be done thorough the mechanism of comparison. Compare the selected facial feature from the images or facial database.

Traditional recognition system identify faces through landmarks or specified features which may lead to fatal recognition but now a days three dimentional techniques are used. These techniques use 3D sensors which can detect images from different angels and are very accurate and precise.

 The Face API provides four recognition functionalities:

      Face verification

      similar face looking out

      Automatic face grouping

      Person identification.

It is widely used in security systems, celebrity recognition and image tagging applications.

Face Verification:

It is also called face authentication. Face API verification will perform AN authentication against 2 detected faces. It concerns with the validation of image. Metric, mlp, gmm and compute_perf are major programs used in verification of faces.
Similar face Searching: 

Similar Face Searching: 

Face API will look for faces supported similarity. By providing one target detected face, and a collection of unknown faces to go looking with, our service will come atiny low set of faces that look most like the target face.

Face Grouping:
Face API will mechanically cluster detected faces supported similarity. This API takes one set of unknown faces, then convert it into many teams. Every cluster could be a disjointed correct set of the initial unknown face set, it'llcontain similar faces which will be thought-about mutually person. This API focus on many points and detect more than one faces.

Face Identification:

Face API identification will determine folks from a detected face. The folks info (defined as someone group) got to be outlined ahead for correct identification.
The following figure is AN example of someone cluster named "MyFriends". every cluster might have up to one thousand folks outlined. Meanwhile, everyone will register one or a lot of faces.

After someone cluster has been created and trained, identification will then be performed against the cluster and a replacement take a look at face. If the face is known as someone outlined within the cluster, the person are going to be came.

Major Role:

It is frequently used in bio metrics, video surveillance, human computer interface and database management system.

It is also used in digital cameras to detect the face.

Weakness:

It does not work well in when the light is poor, sunglasses or when the some object covers the subject face.

Even some facial expression can make it less effective like big smile, long hairs and darker toner skins etc.