Microsoft Azure Computer Vision - Detailed Review
Search Tools
Microsoft Azure Computer Vision - Product Overview
Microsoft Azure Computer Vision
Microsoft Azure Computer Vision is a comprehensive AI-driven service within the Azure AI Services category, designed to help developers and businesses extract valuable insights from images and videos.
Primary Function
The primary function of Azure Computer Vision is to enable applications to analyze and interpret visual data from images and videos. This involves using advanced algorithms to process visual features, extract information, and provide actionable data.
Target Audience
The target audience for Azure Computer Vision includes AI engineers, developers, and businesses looking to integrate computer vision capabilities into their applications and workflows. This service is particularly useful for those with familiarity in Azure and the Azure portal, although it also caters to developers of various skill levels.
Key Features
Image Analysis
Azure Computer Vision can analyze images to detect and classify objects, faces, and adult content. It also generates auto-created text descriptions and captions for images. This service can pull from over 10,000 concepts and objects to provide detailed insights.
Optical Character Recognition (OCR)
The service includes OCR capabilities to extract printed and handwritten text from images, supporting various languages and writing styles. This is achieved through the Read API and OCR API, which can handle text on different surfaces and backgrounds.
Facial Recognition
The Face service within Azure Computer Vision detects, recognizes, and analyzes human faces. It can identify facial landmarks, emotions, and other attributes such as age, facial hair, and glasses. This is useful for applications requiring face-based authentication and other facial analysis tasks.
Object Detection
Azure Computer Vision allows for custom object detection models to be trained and deployed. This involves identifying the location of objects within images along with their classification, which can be achieved with a relatively small set of training images using the Custom Vision Service.
Spatial Analysis
The service includes spatial analysis to understand people’s presence and movements within physical areas in real time. This is particularly useful for tracking movement and analyzing environments through video feeds.
Video Analysis
Azure Computer Vision also supports video-related features such as video retrieval and spatial analysis. This allows for the analysis of video content to detect people, objects, and other visual elements.
By integrating these features, Azure Computer Vision enables developers to build intelligent applications that can interpret and act on visual data, enhancing various business processes and user experiences.
Microsoft Azure Computer Vision - User Interface and Experience
User Interface of Microsoft Azure Computer Vision
The user interface of Microsoft Azure Computer Vision, particularly through the newly introduced Vision Studio, is designed to be user-friendly and accessible, even for those without extensive coding experience.
Key Features of Vision Studio
- UI-Based Tools: Vision Studio offers a set of UI-based tools that allow users to explore, demo, and evaluate various features of Azure Computer Vision without writing any code.
- Try-It-Out Experiences: Users can test different services such as Optical Character Recognition (OCR), Spatial Analysis, Face recognition, and Image Analysis using either their own images or sample images provided by Azure.
Ease of Use
- No-Code Approach: Vision Studio enables users to start experimenting with Azure Computer Vision services quickly, using a no-code approach that provides JSON and text responses. This makes it easy for users to test the features and see the results immediately.
- Simple Setup: To get started, users need an Azure subscription and a resource for Azure AI services. The process involves signing in to Azure, choosing or creating a Vision resource, and selecting a location for the resource. This setup is straightforward and guided through the Vision Studio website.
User Experience
- Intuitive Interface: The interface is intuitive, allowing users to select different features and try them out with minimal steps. For example, users can upload their own images to test OCR, analyze spatial movements in videos, or detect and recognize faces in images.
- Real-Time Feedback: The services in Vision Studio provide real-time feedback, such as extracting text from images, analyzing spatial movements, or detecting objects and faces. This immediate feedback helps users understand the capabilities of the services quickly.
Accessibility
- Sample Images: Users can try out the services using sample images provided by Azure, which is helpful for those who do not have their own images to test with. This feature allows anyone to explore the capabilities of Azure Computer Vision without any initial setup.
- Web-Based: The entire experience is web-based, requiring no additional software installation. This makes it accessible from any device with a web browser, enhancing the overall user experience.
Overall, the user interface of Azure Computer Vision through Vision Studio is designed to be easy to use, intuitive, and accessible, making it a valuable tool for both beginners and experienced users looking to leverage computer vision capabilities.
Microsoft Azure Computer Vision - Key Features and Functionality
Microsoft Azure Computer Vision
Microsoft Azure Computer Vision is a comprehensive suite of tools and services that enable machines to interpret and make decisions based on visual data. Here are the main features and their functionalities:
Azure Cognitive Services: Vision API
The Vision API is a key component of Azure Cognitive Services, providing a range of functionalities such as:
- Image Recognition: The API can analyze images to identify objects, people, and text within them. It can also suggest captions and tags for images.
- Object Detection: This feature identifies objects within an image and delineates their boundaries with bounding boxes. It is useful for applications like automated inventory management.
- Optical Character Recognition (OCR): The API can extract text from images, making it useful for document processing and other text-based applications.
Azure Custom Vision
This service allows developers to create and train their own custom computer vision models. By uploading labeled images, users can train models to recognize specific objects or categories that are relevant to their business needs. This is particularly useful for scenarios where pre-trained models are not sufficient.
Azure Face API
The Face API is part of Azure Cognitive Services and focuses on facial recognition and analysis. It can detect faces in images, estimate age and emotion, and verify if two faces belong to the same person. This API is useful for authentication, sentiment analysis, and user engagement.
Azure Computer Vision SDK
The Azure Computer Vision SDK provides a development framework for integrating Azure’s computer vision services into applications. It supports various programming languages, such as Python and C#, allowing developers to send requests to the Vision API and analyze images seamlessly.
Image Classification
Azure offers services for image classification, which involves categorizing images into predefined classes or categories. Azure Custom Vision is particularly useful here, as it allows users to build and deploy custom image classification models with minimal machine learning expertise.
Semantic Segmentation
This feature involves segmenting images into their constituent parts, assigning a class label to each pixel. It is useful in applications where detailed image analysis is required, such as in medical imaging or autonomous vehicles.
Integration with IoT and Other Services
Azure Computer Vision can be integrated with Internet of Things (IoT) devices, enabling the development of intelligent and responsive systems. This integration allows for real-time image analysis and decision-making in various environments, such as manufacturing and quality control.
Real-World Applications
- Retail and E-Commerce: Used for inventory management, shelf monitoring, and cashier-less checkout systems. Visual search capabilities enable users to find products by uploading images.
- Healthcare: Utilized for medical image analysis to aid in disease diagnosis and patient identification through facial recognition.
- Manufacturing and Quality Control: Employed for automated inspection and quality control processes.
Troubleshooting and Best Practices
To ensure the optimal performance of computer vision models, it is crucial to address challenges such as insufficient data quality and overfitting. Ensuring diverse, representative, and accurately labeled training data is key to mitigating these issues.
These features and functionalities of Azure Computer Vision enable developers and businesses to build innovative applications that can analyze and derive valuable insights from visual data, automate repetitive tasks, and deliver enhanced user experiences.
Microsoft Azure Computer Vision - Performance and Accuracy
Performance Metrics
Azure Computer Vision, including Custom Vision, evaluates performance using metrics such as precision, recall, and mean average precision (mAP). Precision measures the percentage of identified classifications that were correct, while recall measures the percentage of actual classifications that were correctly identified. The mAP is the average value of the average precision, which is the area under the precision/recall curve.
Data Quality and Quantity
The accuracy of Custom Vision models heavily depends on the quality, quantity, and variety of the labeled data provided during training. A balanced dataset with sufficient images for each class is crucial for achieving high performance. For instance, it is recommended to have at least 50 labeled images per tag for classification and 15 for object detection, though more is generally better.
Probability Threshold
The probability threshold is another critical factor that affects the model’s performance. Adjusting this threshold can trade off between precision and recall. A high threshold increases precision but may reduce recall, while a low threshold increases recall but may introduce more false positives. The default threshold is 50%, but it can be adjusted based on the specific needs of the project.
Iterative Training and Testing
Building a Custom Vision model is an iterative process. Each training iteration updates the performance metrics, allowing you to evaluate and improve the model over time. It is recommended to test the model in an isolated environment, gather feedback, and further train the model until it meets the desired performance levels.
Image Specifications
There are specific limits and guidelines for image uploads. Images must be between 256 and 10,240 pixels in height and width, with a maximum aspect ratio of 25:1. Images smaller than 256 pixels will be accepted but upscaled. The maximum file size for training images is 6 MB and 4 MB for prediction images.
Subscription Limits
The performance of Azure Computer Vision can also be influenced by the subscription tier. For example, the free tier (F0) has limitations such as 2 projects, 5,000 training images per project, and 10,000 predictions per month, whereas the standard tier (S0) offers more generous limits, including 100 projects, 100,000 training images per project, and unlimited predictions.
Areas for Improvement
- Data Quality: Ensuring high-quality, diverse, and balanced datasets is essential for improving model accuracy.
- Overfitting: Preventing overfitting by testing the model with additional data and adjusting the model as necessary is crucial.
- Probability Threshold: Adjusting the probability threshold to balance precision and recall according to the project’s specific needs.
- Image Resolution and Quality: Higher resolution images generally produce better results, especially for tasks like OCR. Adjusting image resolution, contrast, and other attributes can improve accuracy.
By considering these factors and best practices, users can optimize the performance and accuracy of Azure Computer Vision models, making them more effective in various search and AI-driven applications.
Microsoft Azure Computer Vision - Pricing and Plans
Pricing Structure for Microsoft Azure Computer Vision
The pricing structure for Microsoft Azure Computer Vision, which includes various APIs and services, can be broken down into several components and plans.
Microsoft Computer Vision API
- This API has a pay-as-you-go model, with a starting price of $1.00 per 1,000 transactions.
- There is a single plan, S1 – Web/Container, priced at $2.50 per 1,000 transactions.
- Additionally, Microsoft Computer Vision API offers a free plan with limited features.
Azure Custom Vision
- Azure Custom Vision allows users to create and train their own computer vision models.
- There are two tiers:
- F0 (Free): This tier includes 2 projects, 5,000 training images per project, 10,000 predictions per month, and other limited features such as 50 tags per project and 20 iterations.
- S0 (Standard): This tier includes 100 projects, 100,000 training images per project, unlimited predictions per month, 500 tags per project, and other enhanced features compared to the free tier.
Azure Face API
- The Face API is part of Azure Cognitive Services and is priced separately.
- It does not have a free tier specifically mentioned, but it follows a pay-as-you-go model, with pricing details available on the Azure pricing page.
Free Azure Services
- Microsoft Azure offers several free services and credits that can be used to try out various Azure products, including some aspects of Azure Computer Vision.
- New Azure customers can get popular services free for 12 months and over 65 other services free always, along with a $200 credit to use in the first 30 days.
Key Features and Limits
- Microsoft Computer Vision API: Includes features like image recognition, object detection, and optical character recognition (OCR).
- Azure Custom Vision: Allows for custom model creation and training, with limits on the number of projects, training images, and predictions based on the tier chosen.
- Azure Face API: Provides facial recognition, age and emotion estimation, and face verification.
Summary
In summary, the pricing for Azure Computer Vision services is based on a pay-as-you-go model, with specific tiers and limits for each service. There are free options available, particularly for Azure Custom Vision and through the general Azure free services and credits.
Microsoft Azure Computer Vision - Integration and Compatibility
Integration with Azure Cognitive Services
Azure Computer Vision is part of Azure Cognitive Services, which includes a range of APIs and services such as the Vision API, Custom Vision, and Face API. These services can be integrated into applications using the Azure Computer Vision SDK, which supports multiple programming languages including Python, C#, and others. This SDK facilitates sending requests to the Vision API, analyzing images, and extracting valuable information.
Compatibility with Azure AI Services
Azure Computer Vision can be combined with other Azure AI services to handle more complex tasks. For example, you can integrate Azure AI Vision with Azure OpenAI models for multimodal scenarios. This integration allows you to use Azure AI Vision for tasks like OCR or image analysis and then pass the extracted data to a GPT-4 model for further reasoning or response generation.
Integration with IoT Devices
Azure Computer Vision services can be seamlessly integrated with Internet of Things (IoT) devices. This integration enables the development of intelligent and responsive systems, where computer vision can be used to analyze data from IoT devices in real-time.
Cross-Platform Development
The Azure Computer Vision SDK supports development on various platforms. Developers can use the SDK to build applications on Windows, Linux, or macOS, and deploy them on cloud, on-premises, or edge environments. This flexibility makes it easier to integrate computer vision capabilities into a wide range of applications.
Real-World Applications
In real-world scenarios, Azure Computer Vision is used across different sectors such as retail, healthcare, and manufacturing. For instance, in retail, it is used for inventory management and visual search; in healthcare, it aids in medical image analysis; and in manufacturing, it is employed for quality control.
Security and Performance
Azure ensures that computer vision applications adhere to security best practices, including encryption, secure communication, and proper access controls. Additionally, Azure provides tools for monitoring and scaling to address performance bottlenecks, ensuring that applications run efficiently and securely.
No-Code and Low-Code Options
For users who are not developers, Azure offers no-code and low-code options through the Power Platform. This allows non-technical individuals to create and integrate AI models, including computer vision, into their applications without extensive coding knowledge.
Overall, Microsoft Azure Computer Vision is highly versatile and compatible with a wide range of tools, services, and platforms, making it a powerful tool for integrating AI capabilities into various applications.
Microsoft Azure Computer Vision - Customer Support and Resources
Support Options in the Azure Portal
To address common issues or seek troubleshooting help, you can use the Azure portal. Here’s how:
- Go to your Azure AI services resource in the Azure portal.
- In the left pane, select Support Troubleshooting under the Help section.
- Describe your issue and answer the remaining questions in the form to find relevant Learn articles and other resources that might help you resolve your problem.
Creating a Support Request
If you need more direct support, you can create a support request in the Azure portal. Here are the steps:
- Go to the New support request page.
- Choose your Issue type and select Cognitive Services in the Service type dropdown field.
- Follow the instructions to submit your request. Azure offers various support plans, including Developer, Standard, and Professional Direct (ProDirect), each with different response times based on the severity of the issue.
Community and Expert Support
In addition to direct support, you can engage with the Azure community and experts:
- Connect with Microsoft engineers and Azure community experts through community support forums.
- Follow @AzureSupport on Twitter for answers and support from Azure experts.
- Utilize real-time dashboards, custom recommendations, and alerts through tools like Azure Service Health and Azure Monitor to manage and optimize your resources.
Documentation and Guides
Microsoft provides comprehensive documentation and guides to help you get started and troubleshoot issues:
- The Azure Cognitive Services documentation includes detailed guides on setting up and using the Vision API, Custom Vision, and other related services.
- You can find step-by-step instructions on creating a computer vision resource, accessing the Vision API, and integrating it into your codebase.
Additional Resources
For further learning and implementation, you can explore:
- GitHub examples and code snippets to see the features in action.
- Free online tools and apps, such as the Free Online Document Redaction App, which can help you perform redactions using Azure Computer Vision.
These resources ensure you have the support and information needed to effectively use and troubleshoot Microsoft Azure Computer Vision services.
Microsoft Azure Computer Vision - Pros and Cons
Advantages of Microsoft Azure Computer Vision
Microsoft Azure Computer Vision offers several significant advantages that make it a powerful tool for various applications:Scalability and Flexibility
Azure Computer Vision allows for dynamic scaling of resources based on workload, ensuring that your application can handle increased demands without compromising performance.Advanced Image Processing
The service includes a range of functionalities such as image recognition, object detection, optical character recognition (OCR), facial recognition, and content moderation. These capabilities enable developers to extract valuable insights from images and videos.Customization
With Azure Custom Vision, developers can create and train their own computer vision models using their specific data. This allows for more accurate and relevant visual recognition tasks.Integration with Other Services
Azure Computer Vision seamlessly integrates with other Azure services, such as IoT devices and edge computing, enabling real-time processing and analysis of visual data. This integration also extends to natural language interactions, enhancing accessibility and driving SEO.Security and Compliance
Azure prioritizes security with features like role-based access control (RBAC) and encryption, ensuring data protection and compliance with privacy regulations.Edge Computing and Real-Time Processing
The integration of computer vision with edge computing enables devices to analyze and respond to visual information locally, reducing latency and improving real-time processing capabilities.Explainable AI
Azure focuses on interpretability and transparency in machine learning models, ensuring that predictions and decisions can be understood and trusted, particularly in critical applications like healthcare.Disadvantages of Microsoft Azure Computer Vision
While Azure Computer Vision offers many benefits, there are also some challenges and limitations to consider:Data Quality Issues
Poor data quality, such as insufficient or inaccurately labeled training data, can significantly hinder the performance of computer vision models. Ensuring diverse, representative, and accurately labeled data is crucial.Overfitting
Models can suffer from overfitting, where they perform well on training data but poorly on new data. Regularization techniques and using more diverse training data can help mitigate this issue.Deployment Challenges
Deploying computer vision models into production can be complex. Careful planning, testing, and utilizing Azure’s deployment tools are essential to avoid issues.Performance Bottlenecks
Identifying and addressing performance bottlenecks is important for achieving optimal results. Monitoring the application’s performance and using Azure’s scaling options can help manage increased workloads.Security Concerns
While Azure provides strong security features, ensuring that the application adheres to security best practices, including encryption and secure communication, is still a top priority.Cost Concerns
Using Azure services, including Azure Computer Vision, can be costly. Businesses need to consider the financial implications and ensure that the benefits justify the expenses.Complexity
Azure’s suite of services can be complex to set up and manage, especially for those without extensive experience in cloud computing and AI. By considering these advantages and disadvantages, you can make an informed decision about whether Microsoft Azure Computer Vision is the right fit for your needs. Microsoft Azure Computer Vision - Comparison with Competitors
Unique Features of Azure Computer Vision
- Comprehensive Image Analysis: Azure Computer Vision offers a wide range of image analysis capabilities, including object detection, face recognition, and optical character recognition (OCR). It can extract various visual features such as objects, faces, adult content, and auto-generated text descriptions from images.
- Custom Vision Service: This service allows users to build, improve, and deploy their own image classifiers and object detection models with minimal AI expertise. It is particularly useful for specific business use cases where pre-existing models may not suffice.
- Face Service: Azure’s Face service provides advanced facial analysis, including face detection, facial landmarks identification, and face recognition. It can also analyze attributes such as age, emotion, and facial hair.
- Integration with Other Azure Services: Azure Computer Vision can be seamlessly integrated with other Azure AI services, such as Azure Cognitive Search, Form Recognizer, and Video Indexer, enhancing its capabilities in various applications.
Potential Alternatives
Google Cloud Vision API
- Similar Capabilities: Google Cloud Vision API also offers image analysis, object detection, face detection, and OCR. It is known for its high accuracy in identifying objects and scenes within images.
- Difference: Google Cloud Vision API might require more technical expertise to customize models compared to Azure’s Custom Vision Service. However, it integrates well with other Google Cloud services like AutoML for custom model building.
Amazon Rekognition
- Similar Capabilities: Amazon Rekognition provides image and video analysis, including object detection, face recognition, and text recognition.
- Difference: Amazon Rekognition is tightly integrated with AWS services, making it a good choice for those already invested in the AWS ecosystem. However, its customization options might not be as user-friendly as Azure’s Custom Vision Service.
IBM Watson Visual Recognition
- Similar Capabilities: IBM Watson Visual Recognition offers image classification, object detection, and face detection.
- Difference: IBM Watson Visual Recognition is part of the broader IBM Watson suite, which can be beneficial for those using other IBM AI services. However, it may not offer the same level of customization and ease of use as Azure’s Custom Vision Service.
Conclusion
Microsoft Azure Computer Vision stands out with its comprehensive suite of services, ease of use, and strong integration with other Azure AI tools. While alternatives like Google Cloud Vision API, Amazon Rekognition, and IBM Watson Visual Recognition offer similar capabilities, Azure’s Custom Vision Service and seamless integration with other Azure services make it a compelling choice for many users. If you are already using Microsoft Azure for other services, Azure Computer Vision’s integration capabilities can be a significant advantage. However, if you are invested in another cloud ecosystem or prefer specific features of another service, those alternatives might be more suitable for your needs. Microsoft Azure Computer Vision - Frequently Asked Questions
What is Azure Computer Vision and what does it do?
Azure Computer Vision is a cloud-based service that provides advanced algorithms for image processing and analysis. It helps applications interpret and understand the content of images, including detecting objects, classifying images, reading text through Optical Character Recognition (OCR), and analyzing faces.
How does Azure Computer Vision analyze images?
Azure Computer Vision analyzes images by converting pixel values into numeric features that are used to train machine learning models. These models can predict human-friendly labels such as objects, attributes, and text, with associated confidence values. The service can extract various visual features, including objects, faces, adult content, and auto-generated text descriptions.
What are the different services offered under Azure AI Vision?
Azure AI Vision includes several services:
- Azure Computer Vision: Uses pre-existing advanced image analysis algorithms.
- Azure Custom Vision: Allows you to build, improve, and deploy your own image classifiers.
- Face Service: Provides advanced algorithms for detecting, recognizing, and analyzing human faces.
- Optical Character Recognition (OCR): Extracts printed and handwritten text from images.
- Video Analysis: Includes features like Spatial Analysis and Video Retrieval to analyze video content.
How much does Azure Computer Vision cost?
The pricing for Azure Computer Vision follows a pay-as-you-go model. The cost starts at $1.00 per 1,000 transactions, with a free tier available that includes limited features. There is also an option to start with a free account that includes $200 credit for the first 30 days.
What image formats and sizes are supported by Azure Computer Vision?
Azure Computer Vision can analyze images in JPEG, PNG, GIF, or BMP format. The file size must be less than 4 megabytes (MB), and the dimensions must be greater than 50 x 50 pixels. For the Read API, the dimensions must be between 50 x 50 and 10,000 x 10,000 pixels.
Does Azure Computer Vision store or use my images for training its models?
No, Microsoft does not store your images or use them to train the underlying models. Images are processed and then deleted after analysis, ensuring your data privacy.
Can I customize the models in Azure Computer Vision?
Yes, you can customize models using the Azure Custom Vision Service. This involves preparing your training image set, uploading the data, training and validating the model, and then publishing it for use. Customization is also possible for other services like the Face Service, where you can fine-tune models for specific scenarios.
What is the availability and reliability of Azure AI Vision services?
Azure AI Vision services guarantee 99.9 percent availability. However, no Service Level Agreement (SLA) is provided for the free pricing tier.
How does the Face Service in Azure AI Vision work?
The Face Service detects, recognizes, and analyzes human faces in images. It can perform face detection, face analysis (including facial landmarks), and face recognition for authentication purposes. The service also returns attributes such as age, emotion, facial hair, and more.
Can I use Azure Computer Vision for video analysis?
Yes, Azure AI Vision includes Video Analysis features such as Spatial Analysis, which analyzes the presence and movement of people in video feeds, and Video Retrieval, which allows you to create an index of videos for search purposes.
Microsoft Azure Computer Vision - Conclusion and Recommendation
Microsoft Azure Computer Vision
Microsoft Azure Computer Vision is a comprehensive and powerful tool within the Azure AI services, offering a wide range of capabilities that can significantly benefit various types of users and organizations.
Key Capabilities
- Image Recognition and Analysis: Azure Computer Vision allows machines to analyze and interpret visual information from images and videos, using deep learning algorithms and neural networks. This includes image recognition, object detection, and optical character recognition (OCR).
- Custom Models: With Azure Custom Vision, developers can create and train their own computer vision models, which is particularly useful for scenarios where pre-trained models are not sufficient.
- Facial Recognition: The Face API enables facial detection, identification, age estimation, and emotion analysis, making it useful for authentication, sentiment analysis, and user engagement.
- Text Extraction: The service can extract text from images and documents using OCR, which is beneficial for automatic indexing and search purposes.
Who Would Benefit Most
- AI Engineers and Developers: These professionals can leverage Azure Computer Vision to integrate AI capabilities into their applications, such as image classification, object detection, and facial recognition.
- Businesses: Companies seeking to automate processes like image analysis, text extraction, and facial recognition can greatly benefit from these services. For example, retail businesses can use it to analyze CCTV surveillance videos to identify and tag human behavior.
- Researchers: Researchers in fields requiring image and video analysis can utilize Azure Computer Vision to extract insights and information from visual data.
Overall Recommendation
Azure Computer Vision is highly recommended for anyone looking to integrate advanced image and video analysis capabilities into their applications. Here are some key reasons:
- Ease of Use: The services are fully cloud-based and require no additional infrastructure, making it easy to get started quickly.
- Customization: The ability to create and train custom models using Azure Custom Vision adds a layer of flexibility and accuracy specific to the user’s needs.
- Comprehensive Features: The suite includes a wide range of functionalities, from image classification and object detection to facial recognition and text extraction, making it a versatile tool for various applications.
In summary, Microsoft Azure Computer Vision is a powerful and versatile tool that can be highly beneficial for a wide range of users, from AI engineers and developers to businesses and researchers, due to its comprehensive features, ease of use, and customization options.