# Computer Vision & Natural Language Processing: Sample Questions and Answers

> **Note**
>
> The questions and answers provided in this study guide are for practice purposes only and are not official practice questions.
> They are intended to help you prepare for the [AI-102 Microsoft certification exam](https://learn.microsoft.com/en-us/credentials/certifications/azure-ai-engineer/?practice-assessment-type=certification).
> For additional preparation materials and the most up-to-date information, please refer to the [official Microsoft documentation](https://learn.microsoft.com/en-us/credentials/certifications/azure-ai-engineer/?practice-assessment-type=certification#certification-prepare-for-the-exam).

> - Implement computer vision solutions <br/> 
> - Implement natural language processing solutions

<details markdown="1">
<summary><b>List of References </b> (Click to expand)</summary>

- [Microsoft Certified: Azure AI Engineer Associate](https://learn.microsoft.com/en-us/credentials/certifications/azure-ai-fundamentals/) - Overview of the AI-102 certification, including skills measured and exam details.
- [Study Guide for Exam AI-102: Designing and Implementing an Azure AI Solution](https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-900) - Detailed study guide with topics covered in the exam and links to additional resources.
- [Course AI-102T00-A: Designing and Implementing an Azure AI Solution](https://learn.microsoft.com/en-us/training/courses/ai-900t00) - Training course that covers designing and implementing AI solutions using Azure AI services.

</details>

<details markdown="1">
<summary><b>List of questions/answers </b> (Click to expand)</summary>

- [Q1: Making a Direct API Request Using Smart Cropping Feature](#q1-making-a-direct-api-request-using-smart-cropping-feature)
- [Q2: Ensuring Completion of OCR Operation Using Computer Vision Client Library](#q2-ensuring-completion-of-ocr-operation-using-computer-vision-client-library)
- [Q3: Managing Access in Language Understanding LUIS](#q3-managing-access-in-language-understanding-luis)
- [Q4: Provisioning a QnA Maker Service](#q4-provisioning-a-qna-maker-service)
- [Q5: Implementing Sentiment Analysis in Customer Feedback System](#q5-implementing-sentiment-analysis-in-customer-feedback-system)
- [Q6: Building a Model for an iOS App to Identify Animals](#q6-building-a-model-for-an-ios-app-to-identify-animals)
- [Q7: Configuring an App for Vision Impaired Users](#q7-configuring-an-app-for-vision-impaired-users)
- [Q8: Training a Custom Form Recognizer Model](#q8-training-a-custom-form-recognizer-model)
- [Q9: Regenerating an API Key](#q9-regenerating-an-api-key)
- [Q10: Creating a Text Analytics Resource](#q10-creating-a-text-analytics-resource)
- [Q11: Deploying Text Analytics API Container](#q11-deploying-text-analytics-api-container)
- [Q12: Updating a Computer Vision Resource](#q12-updating-a-computer-vision-resource)
- [Q13: Language Understanding Model](#q13-language-understanding-model)
- [Q14: Building a Multilingual Chatbot with Sentiment Analysis](#q14-building-a-multilingual-chatbot-with-sentiment-analysis)
- [Q15: Implementing Image Classification with Custom Vision](#q15-implementing-image-classification-with-custom-vision)
- [Q16: Exporting a Custom Vision Model for Offline Use](#q16-exporting-a-custom-vision-model-for-offline-use)
- [Q17: Analyzing Image Using Computer Vision Client Library](#q17-analyzing-image-using-computer-vision-client-library)
- [Q18: Implementing Real-Time Sentiment Analysis in a Customer Feedback System](#q18-implementing-real-time-sentiment-analysis-in-a-customer-feedback-system)
- [Q19: Recognizing Employees' Faces Using the Face Recognition API](#q19-recognizing-employees-faces-using-the-face-recognition-api)

</details>

> **Tip**
>
> **Computer Vision Services:**

| Service | Use Case | Key Features |
|---------|----------|--------------|
| **Computer Vision** | General image analysis | OCR, image analysis, smart cropping, thumbnails |
| **Custom Vision** | Custom image classification/detection | Train custom models, export for offline use |
| **Face** | Face detection and recognition | Face detection, identification, verification |
| **Form Recognizer** | Document processing | Extract text, tables, key-value pairs from forms |

> **Tip**
>
> **Natural Language Processing Services:**

| Service | Use Case | Key Features |
|---------|----------|--------------|
| **Text Analytics** | Text analysis and insights | Sentiment analysis, key phrase extraction, language detection |
| **Language Understanding (LUIS)** | Intent recognition | Build conversational AI, understand user intents |
| **QnA Maker** | Knowledge base chatbots | Create FAQ bots, question-answer pairs |
| **Translator** | Text translation | Multi-language translation, document translation |

> **Tip**
>
> **Computer Vision API Endpoints:**

| Endpoint | Purpose | Parameters |
|----------|---------|------------|
| **analyze** | General image analysis | visualFeatures, details, language |
| **ocr** | Optical character recognition | language, detectOrientation |
| **read** | Modern OCR for text extraction | language (async operation) |
| **generateThumbnail** | Smart cropping thumbnails | width, height, smartCropping |
| **describe** | Image description generation | maxCandidates, language |

> **Tip**
>
> **Text Analytics Capabilities:**

| Feature | Description | Output |
|---------|-------------|--------|
| **Sentiment Analysis** | Determine emotional tone | Positive, Negative, Neutral + confidence |
| **Key Phrase Extraction** | Identify important phrases | List of key phrases |
| **Language Detection** | Identify text language | Language code + confidence |
| **Named Entity Recognition** | Extract entities | People, places, organizations, etc. |

> **Tip**
>
> **Custom Vision Project Types:**

| Project Type | Use Case | Output |
|--------------|----------|--------|
| **Classification** | Categorize entire images | Single or multiple tags per image |
| **Object Detection** | Find and locate objects | Bounding boxes with labels |
| **Multi-class** | One tag per image | Single category assignment |
| **Multi-label** | Multiple tags per image | Multiple category assignments |

> **Tip**
>
> **Custom Vision Domains:**

| Domain | Use Case | Optimization |
|--------|----------|--------------|
| **General** | Wide variety of images | Balanced performance |
| **General (compact)** | Mobile/edge deployment | Smaller model size |
| **Food** | Food and dishes | Food-specific features |
| **Landmarks** | Buildings and monuments | Landmark recognition |
| **Retail** | Products and items | Retail-specific features |

> **Tip**
>
> **Form Recognizer Supported Formats:**

| Format | Max Size | Use Case |
|--------|----------|----------|
| **PDF** | 50 MB | Multi-page documents |
| **JPG/JPEG** | 50 MB | Image documents |
| **PNG** | 50 MB | Image documents |
| **BMP** | 50 MB | Image documents |
| **TIFF** | 50 MB | Multi-page images |

> **Tip**
>
> **Azure AI Services Authentication:**

| Method | Use Case | Security Level |
|--------|----------|----------------|
| **Subscription Key** | Simple API access | Basic |
| **Azure AD Token** | Enterprise applications | High |
| **Managed Identity** | Azure services | Highest |
| **Service Principal** | Application authentication | High |

## Q1: Making a Direct API Request Using Smart Cropping Feature

> You have a Computer Vision resource named contoso1 that is hosted in the West US Azure region. You need to use contoso1 to make a direct API request of a product photo by using the smart cropping feature. How should you complete the API URL? 

```bash
curl -X "Ocp-Apim-Subscription-Key: xxx" \
     -F "image=@sample.png" -H "Content-Type: application/json" \
     https://_________________/vision/v3.1/_________________?width=100&height=100&smartCropping=true"
```

**Value1 Options:**

- [ ] api.projectoxford.ai ❌: `This is incorrect because api.projectoxford.ai is an outdated endpoint.`
- [ ] contoso1.cognitiveservices.azure.com ✅: `This is correct because it specifies the custom subdomain for the Computer Vision resource named contoso1.`
- [ ] westus.api.cognitive.microsoft.com ❌: `This is incorrect because it does not specify the custom subdomain for the Computer Vision resource.`

**Value2 Options:**

- [ ] areaOfInterest ❌: `This is incorrect because areaOfInterest is used to get the raw bounding box coordinates, not for generating a smart-cropped thumbnail.`
- [ ] detect ❌: `This is incorrect because detect is used for detecting objects, not for generating a smart-cropped thumbnail.`
- [ ] generateThumbnail ✅: `This is correct because generateThumbnail is the correct API call for generating a smart-cropped thumbnail.`


## Q2: Ensuring Completion of OCR Operation Using Computer Vision Client Library

> You are developing a method that uses the Computer Vision client library. The method will perform optical character recognition (OCR) on images.

```csharp
public static async Task ReadFileUrl(ComputerVisionClient client, string urlFile)
{
    int numberOfCharsInOperationId = 36;

    var textHeaders = await client.ReadAsync(urlFile, language: "en");
    string operationLocation = textHeaders.OperationLocation;
    Thread.Sleep(2000);
    string operationId = operationLocation.Substring(operationLocation.Length - numberOfCharsInOperationId);

    ReadOperationResult results;
    do
    {
        results = await client.GetReadResultAsync(Guid.Parse(operationId));
        Console.WriteLine("Server status: {0}", results.Status);
     }
     while ((results.Status == OperationStatusCodes.Running || results.Status == OperationStatusCodes.NotStarted));

     var textUrlFileResults = results.AnalyzeResult.ReadResults;
     foreach (ReadResult page in textUrlFileResults)
     {
         foreach (Line line in page.Lines)
         {
             Console.WriteLine(line.Text);
         }
      }
}
```

> During testing, you discover that the call to the GetReadResultAsync method occurs before the read operation is complete. You need to prevent the GetReadResultAsync method from proceeding until the read operation is complete. Modify the code so that it waits until each read result appears as part of the solution.

- [ ] Remove Guid.Parse(operationId) parameter. ❌: `This is incorrect because removing the Guid.Parse(operationId) parameter does not address the issue of waiting for the read operation to complete.`
- [ ] Add code to display the results.Status value. ✅: `This is correct because displaying the results status value helps monitor the progress of the read operation, but it does not alone ensure waiting for completion.`
- [ ] Add code to check for null values. ❌: `This is incorrect because checking for null values does not address the issue of waiting for the read operation to complete.`
- [ ] Add a call to GetReadResultAsync within a loop that contains a delay. ✅: `This is correct because adding a call to GetReadResultAsync within a loop that contains a delay ensures that the method waits for the read operation to complete before proceeding.`


## Q3: Managing Access in Language Understanding (LUIS)

> You are building a language model by using a Language Understanding (classic) service. You create a new Language Understanding (classic) resource. You need to add more contributors. What should you use?

- [ ] a conditional access policy in Azure Active Directory (Azure AD) ❌: `This is incorrect because a conditional access policy is used to control access to resources based on conditions, not for adding contributors.`
- [ ] the Access control (IAM) page for the authoring resources in the Azure portal ✅: `This is correct because the Access control (IAM) page for the authoring resources is where you can manage roles and add contributors to the Language Understanding (classic) resource.`
- [ ] the Access control (IAM) page for the prediction resources in the Azure portal ❌: `This is incorrect because the Access control (IAM) page for the prediction resources is used for managing access to prediction resources, not for adding contributors to the authoring resources.`

## Q4: Provisioning a QnA Maker Service

> You plan to provision a QnA Maker service in a new resource group named RG1. In RG1, you create an `App Service plan` named AP1. Which two Azure resources are automatically created in RG1 when you provision the QnA Maker service? 

- [ ] Language Understanding ❌: `This is incorrect because Language Understanding (LUIS) is not automatically created when provisioning a QnA Maker service.`
- [ ] Azure SQL Database ❌: `This is incorrect because an Azure SQL Database is not automatically created when provisioning a QnA Maker service.`
- [ ] Azure Storage ❌: `This is incorrect because Azure Storage is not automatically created when provisioning a QnA Maker service.`
- [ ] Azure AI Search ✅: `This is correct because Azure AI Search is automatically created to index and search the QnA Maker knowledge base.`
- [ ] Azure App Service ✅: `This is correct because an Azure App Service is automatically created to host the QnA Maker service.`

## Q5: Implementing Sentiment Analysis in Customer Feedback System

> You are developing a customer feedback system that will analyze text feedback from users. You need to implement sentiment analysis to determine the overall sentiment of the feedback. The solution must minimize development effort and provide accurate sentiment scores. Which Azure AI Service should you use for each requirement? 

To analyze the sentiment of text feedback: ___________________________

- [ ] Speech ❌: `This is incorrect because the Speech service is used for processing and recognizing speech, not for analyzing text sentiment.`
- [ ] Face ❌: `This is incorrect because the Face service is used for detecting and recognizing faces, not for analyzing text sentiment.`
- [ ] Text Analytics ✅: `This is correct because the Text Analytics service provides sentiment analysis capabilities to determine the overall sentiment of text feedback.`

To extract key phrases from the text feedback: ___________________________

- [ ] Speech ❌: `This is incorrect because the Speech service is used for processing and recognizing speech, not for extracting key phrases from text.`
- [ ] Text Analytics ✅: `This is correct because the Text Analytics service provides key phrase extraction capabilities to identify important phrases in the text feedback.`
- [ ] Form Recognizer ❌: `This is incorrect because Form Recognizer is used for extracting information from forms and documents, not for extracting key phrases from text.`

To detect the language of the text feedback: ___________________________

- [ ] Custom Vision ❌: `This is incorrect because Custom Vision is used for image classification and object detection, not for detecting the language of text.`
- [ ] Speech ❌: `This is incorrect because the Speech service is used for processing and recognizing speech, not for detecting the language of text.`
- [ ] Text Analytics ✅: `This is correct because the Text Analytics service provides language detection capabilities to identify the language of the text feedback.`

## Q6: Building a Model for an iOS App to Identify Animals

> You are building a model that will be used in an iOS app. You have images of various animals, including cats, dogs, and birds. Each image contains either a cat, a dog, or a bird. You need to use the Custom Vision service to detect which animal is in the image. How should you configure the project in the Custom Vision portal?

**Project Types:** ___________________________

- [ ] Classification ✅: `This is correct because classification is used to categorize images into predefined classes, such as cats, dogs, and birds.`
- [ ] Object Detection ❌: `This is incorrect because object detection is used to identify and locate objects within an image, not to classify the entire image.`

**Classification Types:** ___________________________

- [ ] Multiclass (Single tag per image) ✅: `This is correct because multiclass classification is used when each image contains only one type of animal, such as a cat, dog, or bird.`
- [ ] Multilabel (Multiple tags per image) ❌: `This is incorrect because multilabel classification is used when an image can contain multiple tags, which is not the case here.`

**Domains:** ___________________________

- [ ] Audit ❌: `This is incorrect because the audit domain is not relevant for classifying images of animals.`
- [ ] Food ❌: `This is incorrect because the food domain is not relevant for classifying images of animals.`
- [ ] General ✅: `This is correct because the general domain is suitable for a wide range of image classification tasks, including identifying animals.`
- [ ] General (compact) ✅: `This is correct because the general (compact) domain is optimized for mobile and edge devices, making it suitable for an iOS app.`
- [ ] Landmarks ❌: `This is incorrect because the landmarks domain is used for identifying landmarks, not animals.`
- [ ] Landmarks (compact) ❌: `This is incorrect because the landmarks (compact) domain is used for identifying landmarks, not animals.`
- [ ] Retail ❌: `This is incorrect because the retail domain is used for identifying retail products, not animals.`
- [ ] Retail (compact) ❌: `This is incorrect because the retail (compact) domain is used for identifying retail products, not animals.`

## Q7: Configuring an App for Vision Impaired Users

> You have an app that analyzes images by using the Computer Vision API. You need to configure the app to provide an output for users who are vision impaired. The solution must provide the output in complete sentences. Which API call should you perform?

- [ ] readInStreamAsync ❌: `This is incorrect because readInStreamAsync is used for reading text from images, not for providing descriptive sentences.`
- [ ] tagImageInStreamAsync ❌: `This is incorrect because tagImageInStreamAsync is used for tagging images with keywords, not for providing descriptive sentences.`
- [ ] describeImageInStreamAsync ✅: `This is correct because describeImageInStreamAsync is used to generate a description of the image content in complete sentences, which is suitable for users who are vision impaired.`
- [ ] analyzeImagesByDomainInStreamAsync ❌: `This is incorrect because analyzeImagesByDomainInStreamAsync is used for analyzing images within a specific domain, not for providing descriptive sentences.`

## Q8: Training a Custom Form Recognizer Model

> You build a custom Form Recognizer model. You receive sample files to use for training the model as shown in the following table.

| Name  | Type | Size  |
|-------|------|-------|
| File1 | PDF  | 30 MB |
| File2 | MP4  | 150 MB|
| File3 | JPG  | 40 MB |
| File4 | PDF  | 170 MB|
| File5 | GIF  | 15 MB |
| File6 | JPG  | 30 MB |

> Which three files can you use to train the model?

- [ ]  File1 ✅: `This is correct because it is a PDF file within the size limit for training, which is 50 MB for PDF files.`
- [ ] File2 ❌: `This is incorrect because MP4 files are not supported for training a Form Recognizer model`
- [ ]  File3 ✅: `This is correct because it is a JPG file within the size limit for training, which is 50 MB for image files.`
- [ ] File4 ❌: `This is incorrect because it is a PDF file that exceeds the size limit for training, which is 50 MB for PDF files.`
- [ ] File5 ❌: `This is incorrect because GIF files are not supported for training a Form Recognizer model`
- [ ]  File6 ✅: `This is correct because it is a JPG file within the size limit for training, which is 50 MB for image files.`

## Q9: Regenerating an API Key

> You successfully run the following HTTP request.

```http
POST https://management.azure.com/subscriptions/18e51a87-3a69-47a8-aedc-a5474517f08a/resourceGroups/RGI/providers/Microsoft.AIServices/accounts/conosci1/regenerateKey?api-version=2017-04-18

Body {"keyName": "Key2"}
```

> What is the result of the request?

- [ ] `A new query key was generated.` ❌: `This is incorrect because the request regenerates a subscription key, not a query key.`
- [ ] `A key for Azure AI Services was generated in Azure Key Vault.` ❌: `This is incorrect because the request regenerates a key for the AI Services account, not in Azure Key Vault.`
- [ ] `The primary subscription key and the secondary subscription key were rotated.` ❌: `This is incorrect because the request regenerates only one key, not both.`
- [ ]  `The secondary subscription key was reset.` ✅: `This is correct because the request regenerates the secondary subscription key specified by "Key2".`

## Q10: Creating a Text Analytics Resource

> You have the following C# method for creating Azure AI Services resources programmatically.

```csharp
static void create_resource(AIServicesManagementClient client, string resource_name, string kind, string account_tier, string location) 
{
    AIServicesAccount parameters = new AIServicesAccount(null, null, kind, location, resource_name,
        new AIServicesAccountProperties(), new Sku(account_tier));
    client.AIServicesAccounts.create(resource_group_name, account_tier,
        parameters);
}
```

> You need to call the method to create a Text Analytics resource in the West US Azure region. The resource will be used for sentiment analysis. Which choice should you make?

- [ ]  `create_resource(client, "test", "TextAnalytics", "S0", "westus")` ✅: `This is correct because it specifies the Text Analytics resource kind, which is used for sentiment analysis.`
- [ ] `create_resource(client, "test", "ComputerVision", "F0", "westus")` ❌: `This is incorrect because it specifies the Computer Vision resource kind, not the Text Analytics resource kind.`
- [ ] `create_resource(client, "test", "CustomVision.Training", "S0", "westus")` ❌: `This is incorrect because it specifies the Custom Vision Training resource kind, not the Text Analytics resource kind.`
- [ ] `create_resource(client, "test", "CustomVision.Prediction", "S0", "westus")` ❌: `This is incorrect because it specifies the Custom Vision Prediction resource kind, not the Text Analytics resource kind.`

## Q11: Deploying Text Analytics API Container

> You plan to deploy a containerized version of an Azure AI Services service that will be used for text analysis. You configure `https://contoso.AIservices.azure.com` as the endpoint URL for the service, and you pull the latest version of the Text Analytics API container image. You need to run the container on an Azure virtual machine by using Docker. You configure the command to `/text/answer`.

```bash
docker run --rm -it -p 5000:5000 --memory 8g --cpus 1 \
__________________________ \
-e Eula=accept \
-e Billing=________________ \
-e ApiKey=xxxxxxxxxxxxxxxxxxxxxx
```

**Options for the Container Image:**

- [ ] mcr.microsoft.com/azure-AI-services/textanalytics/keyphrase ❌: `This is incorrect because it specifies the container image for the keyphrase extraction feature, not the sentiment analysis feature.`
- [ ]  mcr.microsoft.com/azure-AI-services/textanalytics/sentiment ✅: `This is correct because it specifies the container image for the sentiment analysis feature.`

**Options for Billing:**

- [ ] http://contoso.blob.core.windows.net ❌: `This is incorrect because it specifies a general Azure Blob Storage endpoint, not the specific billing endpoint for the service.`
- [ ]  https://contoso.AIservices.azure.com ✅: `This is correct because it specifies the endpoint URL for the AI Services, which can be used as the billing endpoint.`

## Q12: Updating a Computer Vision Resource

> Your company has an existing Azure Computer Vision resource that needs to be updated to include a new feature for analyzing handwritten text. The solution must:
> - Apply partial updates to the existing resource. <br/> 
> - Ensure the resource remains in the same region.  <br/> 
> - Use the appropriate HTTP method for partial updates.  <br/> 
> How should you complete the HTTP request to update the resource? 

```http
__________ https://management.azure.com/subscriptions/xxxxxxxx-xxxx-xxxx-xxxx/resourceGroups/RG3/providers/Microsoft.AIServices/accounts/CV1?api-version=2017-04-18

{
  "properties": {
    "features": [
      "__________"
    ]
  }
}
```

- [ ] POST ❌: `This is incorrect because POST is typically used to create a new resource without specifying the resource identifier.`
- [ ] PUT ❌: `This is incorrect because PUT is used to create or update a resource at a specified URI, ensuring idempotency.`
- [ ] AIServices ❌: `This is incorrect because it specifies the general Azure AI Services resource, rather than the specific Computer Vision service.`
- [ ] TextAnalytics ❌: `This is incorrect because it specifies a different service within Azure AI Services, not related to computer vision.`
- [ ]  PATCH ✅: `This is correct because PATCH is used to apply partial updates to an existing resource.`
- [ ]  ComputerVision ✅: `This is correct because it specifies the specific Azure AI Services resource for computer vision tasks.`

```http
PATCH https://management.azure.com/subscriptions/xxxxxxxx-xxxx-xxxx-xxxx/resourceGroups/RG3/providers/Microsoft.AIServices/accounts/CV1?api-version=2017-04-18

{
  "properties": {
    "features": [
      "ComputerVision"
    ]
  }
}
```
## Q13: Language Understanding Model

> You've got 4659 chatbots, each with its own Language Understanding model.
> Often, you need to add the same phrases to all the models.
> You need to update the Language Understanding models programmatically to include these phrases.
> How should you complete the code? 
 
```csharp
var result = await client.Features.<value>(
    appId, versionId, new <value>
    {
        EnabledForAllModels = false,
        IsExchangeable = true,
        Name = "PL1",
        Phrases = "item1,item2,item3,item4,item5"
    });
```

Pick the right values for the correct targets. Each value may be used once, more than once, or not at all.

- [ ] **Phraselist** ❌: `This is incorrect because it is not an action. Represents a collection of phrases used in a language understanding model.`
- [ ] **Phrases** ❌: `This is incorrect because it refers to the content, not an action or method. The actual phrases included in the phrase list.`
- [ ] **SavePhraselistAsync** ❌: `This is incorrect because it is not relevant to the code snippet. Saves the phrase list asynchronously to the language understanding model.`
- [ ] **UploadPhraseListAsync** ❌: `This is incorrect because it is not relevant to the code snippet. Uploads a phrase list asynchronously to the language understanding model.`
- [ ]  **PhraseListCreateObject** ✅: `This is correct because it represents an object for creating a phrase list.`
- [ ]  **AddPhraseListAsync** ✅: `This is correct because it adds a phrase list asynchronously.`

```csharp
var result = await client.Features.AddPhraseListAsync(
    appId, versionId, new PhraseListCreateObject 
    {
        EnabledForAllModels = false,
        IsExchangeable = true,
        Name = "PL1",
        Phrases = "item1,item2,item3,item4,item5"
    });
```

## Q14: Building a Multilingual Chatbot with Sentiment Analysis

> You are tasked with building a chatbot that must: <br/>
> - Handle chit-chat, knowledge base queries, and support multiple languages <br/> 
> - Analyze user messages to determine their emotional tone (sentiment analysis) <br/>
> - Automatically choose the most appropriate language model <br/>
> Which services should you integrate into the chatbot?

- [ ] **Translator, Speech, and Dispatch** ❌: `This is incorrect because it does not include chit-chat or knowledge base support.`
- [ ] **QnA Maker, Language Understanding, and Dispatch** ❌: `This is incorrect because it does not include sentiment analysis.` 
- [ ] **Text Analytics, Translator, and Dispatch** ❌: `This is incorrect because it does not include chit-chat or knowledge base support.`
- [ ]  **Language Understanding, Text Analytics, and QnA Maker** ✅: `This is correct because it supports chit-chat, knowledge base, multilingual models, and performs sentiment analysis.`

## Q15: Implementing Image Classification with Custom Vision

> You are developing an application that needs to classify images into different categories. You plan to use Azure Custom Vision to achieve this. Which steps should you follow to create and train a custom image classification model?

- [ ] Create a new Azure SQL Database. ❌: `This is incorrect because creating an Azure SQL Database is not required for setting up a Custom Vision project.`
- [ ] Configure a virtual network. ❌: `This is incorrect because configuring a virtual network is not required for setting up a Custom Vision project.`
- [ ] Create a new Custom Vision project. 1️⃣: `This is correct because creating a new Custom Vision project is the first step in setting up your image classification model.`
- [ ] Upload images and label them. 2️⃣: `This is correct because uploading and labeling images is necessary for training the model with the correct categories.`
- [ ] Train the model. 3️⃣: `This is correct because training the model on the labeled images is essential for it to learn how to classify new images.`
- [ ] Publish the model. 4️⃣: `This is correct because publishing the model makes it available for use in your application.`

## Q16: Exporting a Custom Vision Model for Offline Use

> You have a Custom Vision service project that performs object detection. The project uses the General domain for classification and contains a trained model. You need to export the model for use on a network that is disconnected from the internet. Which three actions should you perform in sequence?

- [ ] Change the classification type. ❌: `This is incorrect because changing the classification type is not necessary for exporting the model for offline use.`
- [ ] Export the model. 3️⃣: `This is correct because exporting the model is the final step after ensuring it is in the correct domain and retrained.`
- [ ] Retrain the model. 2️⃣: `This is correct because after changing the domain to a compact domain, you need to retrain the model to optimize it for offline use.`
- [ ] Change Domains to General (compact). 1️⃣: `This is correct because you need to change the domain to a compact domain to make the model suitable for offline use.`
- [ ] Create a new classification model. ❌: `This is incorrect because creating a new classification model is not necessary for exporting the existing model for offline use.`


## Q17: Analyzing Image Using Computer Vision Client Library

> You are developing an application that will use the Computer Vision client library. The application has the following code:

```csharp
public async Task AnalyzeImage(ComputerVisionClient client, string localImage)
{
    List<VisualFeatureTypes?> features = new List<VisualFeatureTypes?>
    {
        VisualFeatureTypes.Description,
        VisualFeatureTypes.Tags,
        VisualFeatureTypes.Categories
    };

    try
    {
        ImageAnalysis results = await client.AnalyzeImageInStreamAsync(localImage, features);

        Console.WriteLine(results.Description.Captions[0].Text);
        
        foreach (var tag in results.Description.Captions)
        {
            Console.WriteLine($"\t{tag.Name} ({tag.Confidence:P})");
        }

        foreach (var tag in results.Tags)
        {
            Console.WriteLine($"\t{tag.Name} ({tag.Confidence:P})");
        }
    }
    catch (Exception ex)
    {
         Console.WriteLine(ex.Message);
     }
}
```

For each of the following statements, select if the statement is true.

- [ ] The code will perform face recognition. ❌ `This is incorrect because the code does not include the VisualFeatureTypes.Faces feature, which is required for face recognition.`
- [ ] The code will list tags for objects found in images. ✅ `This is correct because the code includes the VisualFeatureTypes.Tags feature, which lists tags for objects found in images.`
- [ ] The code will read a file from the local file system. ❌ `This is incorrect because the code uses AnalyzeImageInStreamAsync, which analyzes an image stream, not a file from the local file system.`


## Q18: Implementing Real-Time Sentiment Analysis in a Customer Feedback System

> You are developing a customer feedback system that needs to analyze text feedback from users in real-time. The system must provide sentiment analysis to determine the overall sentiment of the feedback. Which Azure Cognitive Service should you use, and what method should you call to achieve this?

**Azure Cognitive Service:** ___________________________

- [ ] Translator Text ❌: `This is incorrect because Translator Text is used for translating text between languages, not for sentiment analysis.`
- [ ] Speech to Text ❌: `This is incorrect because Speech to Text converts spoken language into text, not for analyzing the sentiment of text.`
- [ ] Text Analytics ✅: `This is correct because the Text Analytics service provides sentiment analysis capabilities to determine the overall sentiment of the text feedback.`

**Method:** ___________________________

- [ ] TranslateText ❌: `This is incorrect because TranslateText is used for translating text between languages, not for sentiment analysis.`
- [ ] RecognizeOnceAsync ❌: `This is incorrect because RecognizeOnceAsync is used for recognizing speech, not for analyzing the sentiment of text.`
- [ ] AnalyzeSentiment ✅: `This is correct because the AnalyzeSentiment method is used to analyze the sentiment of the input text, determining whether it is positive, negative, or neutral.`

## Q19: Recognizing Employees' Faces Using the Face Recognition API

> You are developing an application to recognize employees' faces by using the Face Recognition API. Images of the faces will be accessible from a URI endpoint. The application has the following code.

```python
def add_face(subscription_key, person_group_id, person_id, image_uri):
    headers = {
        'Content-Type': 'application/json',
        'Ocp-Apim-Subscription-Key': subscription_key,
    }
    body = {
        "url": image_uri
    }
    conn = http.client.HTTPSConnection('westus.api.cognitive.microsoft.com')
    conn.request("POST", f"/face/v1.0/persongroups/{person_group_id}/persons/{person_id}/persistedFaces", "{body}", headers)
    response = conn.getresponse()
    return response.status
```

For each of the following statements, select if the statement is true. 

- [ ] The code will add a face image to a person object in a person group. ✅ `This is correct because the code sends a POST request to add a face image to a person object in a person group.`
- [ ] The code will work for up to 10,000 people. ❌ `This is incorrect because the Face API supports up to 1,000,000 people per person group.`
- [ ] `add_face` can be called multiple times to add multiple face images to a person. ✅ `This is correct because the `add_face` function can be called multiple times to add multiple face images to a person.`