RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

Microsoft Cognitive Services

Product
The name of the base system (platform): Artificial intelligence (AI, Artificial intelligence, AI)
Developers: Microsoft
Last Release Date: 2017/01/02
Technology: PaaS - Platform As A Service - the Business platform as service,  Speech technologies,  the Systems of video analytics,  Development tools of applications

For April 17, 2017 Cognitive Services (before Project Oxford) is a set of API interfaces by means of which developers can create intellectual applications with support of images and speeches, different languages, technologies of processing of knowledge and search.

As a part of set:

  • The API interface for face recognition - determination, the analysis, grouping of persons in photos and assignment of tags by it
  • Content Moderator is automatic moderation of images, the text and video
  • API of recognition of emotions is personalisation of user interaction by means of recognition of emotions
  • API of computer vision is extraction of important information from images.


The API interface for face recognition

Will recognize on images of the person on front rectangles and front attributes, including external signs, a pose and forecasting of front attributes on the basis of machine learning.

Detection of persons, (2017)
  • Correlation of persons
    • Checks whether two portraits belong to one person, with reliability degree assessment.

  • Search of similar persons

    • the Request allows to find according to the image of the person similar persons from a collection of persons.

  • Grouping of persons

    • Integrates persons in groups on the basis of visual similarity.

  • Face recognition

    • Defines what user possesses the person from a request, on the basis of user-provided data on people and their persons.

API of moderation of images

Automatically estimates existence of offensive and undesirable contents on images of different types, including on the changed images. Text recognition (OCR) and detection of persons increasing the probability of identification of undesirable and speculative contents are provided in Content Moderator. The technology allows to reveal a fuzzy matching to the configured lists of the prohibited and permitted data.

API of moderation of the text

Automatically detects offensive language in more than 100 languages and compares the text with the configured lists. Content Moderator also checks availability of malware, the phishing URL addresses and also use of personal data.


API of moderation of video

Provides anticipatory detection of contents for adults on video thanks to cloud moderation in Azure multimedia services.


Means of the user check

Special means allow the working groups to check results of the automated processing, to approve and change tags for their redefinition.


API of recognition of emotions

This API helps to create the applications reacting to mood, distinguishing emotions and adapting under the person. On a look this cloud API can distinguish satisfaction, lack of emotions, grief, contempt, anger, disgust, fear and surprise. Algorithms of artificial intelligence define these emotions on the basis of the mimic templates applicable even to different cultures.

Recognition of emotions, (2017)


API of computer vision

  • Analysis of the image - Helps to obtain information on visual contents of images. Adds the tags, descriptions and models intended for certain spheres for faultless determination and designation of contents. It is used with parameters for determination of obscene materials and contents for adults, including automatic restriction of such contents. Defines types of images and color schemes in photos.
  • Creation of sketches - creates convenient sketches in storage from any entrance images. Function of creation of sketches changes images according to requirements to the size, a form and style. Applies intellectual cutting to creation of the sketches different from source images proportions, but containing the interesting area.
  • Reading of the text on images - function of optical character recognition allows to reveal the text on images and to take recognized words in the symbol stream suitable for machine reading.
  • Recognition of celebrities - the Celebrities model — one of examples of the models intended for certain spheres. Function of recognition of celebrities can recognize 200 thousand famous figures of business, policy, sport and show business of the whole world.

Text Analytics API

This API allows to analyze emotional coloring of the narration on a scale from 0% (very negative) up to 100% (very positive). Service supports Russian and English and also 13 more languages: French, German, Italian, Spanish, Danish, Dutch, Finnish, Greek, Norwegian, Polish, Portuguese, Swedish and Turkish.

According to Microsoft, scope of technology can be very wide. The solution will be useful, for example, for representatives of business for identification of a ratio of positive and negative reviews about the company, or to communication specialists for tracking of tonality of publications.

Use model: Text Analytics API can be integrated, for example, with Twitter, and then to visualize results of the emotional analysis of tweets using Microsoft Power BI in service of automation Microsoft Flow.

It is possible to try Text Analytics API in operation in a demo.

Robotics