Azure Cognitive Services | API - Datacloudy

 

In this blog we are going to see about Azure Cognitive Service in detail manner. This is one of the important topics that is asked in AI -900 Certification Exam. So i will try to explain the concepts in crisp and clear manner in this blog such that you can remember the concepts forever with the thumb rules. Lets, get started.

In an overview, we can say that Azure Cognitive Service comprise of Computer vision, Optical Character Recognition, Face API, Form recognition, Natural Language Process. Let us see all these topics one by one.

Before going in to the deep dive of the above topics, to achieve this in Azure we can create a Multiple Service Account in Cognitive Service to play with multiple APIs. Then we can go to each documentation and we can give the Endpoint and Key to perform the respective API. And this process is almost same for all the API's mentioned above. Now we can see about each and every API.

i) Computer Vision:

        The key points that is need to be remembered is,

• It Analyze and describe the images
  
  
• Reads text in imagery
  
  
• Read handwriting in imagery
  
  
• Recognize Celebrities and landmarks. 

ii) OCR:

        OCR stands for Optical Character Recognition. It detects the text in an image. To perform Optical Character Recognition on complex documents there is a special option available is called "READ"

iii) Face API:

        It is called Advance Face Detection. That is it identifies the Age, Emotion, glasses etc in a face present in an image. The recommendation of the image is 200X200 pixel or bigger. To identify the person, if we give images up to one thousand faces then it is a "Face List". If up to one million faces then it is "Large Face List". The Functions of Face API are follows,

• Detect:
  
       Which means it detects the face.
  
  
• Find Similar:
  
       Here the output is the array of most similar images.
  
  
• Group :
  
       It divides candidate faces into groups based on face similarity.
  
  
• Identity:
  
       One to many identification, that is closest match for the specific person ranked by confidence level.
  
  
• Verify:
  
       It identifies that do two faces belong to same person

iv) Form Recognition:

        Suppose if we need extract text, key value pair, tables from a document then we need to go for form recognition API in Azure. It has two models, they are Custom Model, Pre-build receipt Model.

v) NLP:

        NLP stands for Natural Language Processing, it is used to get intelligence from conversation, speech or written text in human language. We can consider below subdivision,

• Text Analytics API:
  
      The key feature of this API are,
  
       i) Sentimental Analysis from text. The value ranges between 0.1 to 0.9 . That is when the output value is nearby 0.1 then it is more possibility is Negative sentiment. And similarly if it is near  by 0.9 then it is a positive sentiment.
  
      ii) Identifies the language name
  
      iii) It gives the key phrase of the text. That is it gives the summary. It is awesome right!!!
  
      iv) Identifies the entities containing personal Information.
  
  
• Translator and Speech API:
  
      The key feature of Translator and Speech API are,
  
      i) Converts one language to multiple language . (That is it is a text to text conversion process)
  
      ii) Converts speech to Text. That is Transcription.
  
      iii) Convert Text to Speech.
  
      iv) Perform Translation in Speech as well.
  
  
• Conversational AI:
  
     In Conversational AI we must know two topics,
  
        i) QNA maker:
  
      QNA stands for Question and answer, It converts Frequently Asked Question (FAQ) to QNA Bot. Next to this we have Azure Bot Service this enable QNA to multiple channels like Alexa, Fb, MS office. Thus we attain a Chat bot
  
        ii) LUIS :
  
      LUIS stands for Language Understanding Intelligent Services. It understand spoken and text commands and helps for performing some tasks. It converts the commands into "intent and entities"
  
  For Example: When we say "Order me 2 pizzas" this will be converted into command for the downstream process to order 2 pizzas.

Thus in this blog we saw about Azure Cognitive Service in detail and this will be surely be helpful in AI-900 Exam. Hope this information is helpful.

Thank You!!!