Azure AI Vision
Discover computer vision insights from image and video analysis with OCR and AI
Overview
Enhance your apps with Azure AI Vision
- Azure AI Vision is a unified service that offers innovative computer vision capabilities. Give your apps the ability to analyze images, read text, and detect faces with prebuilt image tagging, text extraction with optical character recognition (OCR), and responsible facial recognition. Incorporate vision features into your projects with no machine learning experience required.
- Automatically caption images with natural language, use smart crop, and classify images (in preview).
- Track movement and analyze environments in real time using computer vision with image analysis and object detection.
- Extract printed and handwritten text from images with mixed languages and writing styles using OCR technology.
- Create apps with facial recognition for a seamless and highly secure user experience.
- Customize image classification and object detection to fit your needs with just a handful of images and without compromising accuracy (in preview).
- Get clear guidance on how to use computer AI Vision responsibly to meet your goals and achieve accurate results.
Features
Analyze visual content in different ways with Azure AI Vision
Image analysis
Image analysis that pulls from more than 10,000 concepts and objects to detect, classify, caption, and generate insights.
Spatial analysis
Spatial analysis to understand people's presence and movements within physical areas in real time.
Optical character recognition (OCR)
Optical character recognition (OCR) to extract printed and handwritten text from images with varied languages and writing styles.
Facial recognition
Facial recognition to create intelligent applications that recognize and verify human identity.
Security
Built-in security and compliance
Microsoft has committed to investing $20 billion in cybersecurity over five years.
We employ more than 8,500 security and threat intelligence experts across 77 countries.
Azure has one of the largest compliance certification portfolios in the industry.
Pricing
Azure AI Vision pricing
Pay for only what you use with no upfront costs. Azure AI Vision uses a pay-as-you-go consumption model based on number of transactions. Learn more about pricing for Azure AI Vision and Face API.
Customer stories
Trusted across industries, by companies of all sizes
Frequently asked questions
- Azure AI Vision and other Azure AI services guarantee 99.9% availability. No service level agreement (SLA) is provided for the free pricing tier.
- No. Microsoft automatically deletes your images and videos after processing and does not train on your data to enhance the underlying models. Video data does not leave your premises, and video data is not stored on the edge where the container runs. Learn more about privacy and terms of usage.
- No, spatial analysis detects and locates human presence in video footage and outputs a bounding box around each person detected. The AI models do not detect faces nor determine individuals’ identities nor demographics.
- The spatial analysis AI models detect and track movements in the video feed based on algorithms that identify the presence of one or more humans by a body bounding box. For each person and bounding box detected in a zone in the camera field of view, the AI models output event data including bounding box coordinates of a person’s body, event type (for example, zone entry or exit, or directional line crossing), pseudonymous identifiers to track the bounding box, and a detection confidence score. This event data is sent to your own instance of Azure IoT Hub.
- Yes. Because model customization is designed to be fine-tuned for your scenario, you need to provide labeled data to train your model.
- The model customization feature of the service is optimized to quickly recognize major differences between images, so you can start prototyping your model with a small amount of data. You may start with as little as one image per label. If you have more labeled images, you may add more. Depending on the complexity of the problem and degree of accuracy required, you can continue adding additional images per label to improve your model.
- It’s both. You can use the site to access a graphical interface for managing datasets, training, and evaluation of models for a no-code experience. Or, as an alternative, you can use the AI Vision APIs.
- You can label the images in Azure Machine Learning studio, which is integrated with Vision Studio for easy export of labeled data. You can also label the data in the Common Objects in Context (COCO) file format and import the COCO file directly into Vision Studio. Vision Studio is a set of UI-based tools that let you explore, build, and integrate features from Azure AI Vision.
- The model customization feature for Azure AI Vision is the next generation of Custom Vision, with improved accuracy and few-shot learning capabilities. You may continue to use Custom Vision, or you can migrate your training data to retrain your model with model customization from Azure AI Vision.
- After using Azure AI Vision to extract insights and text from images and video, you can use text analytics to analyze sentiment, Azure AI Translator to translate text into your desired language, or Immersive Reader to read the text aloud, making it more accessible. Related services and capabilities include Azure AI Document Intelligence to extract key-value pairs and tables from documents, Azure AI Video Indexer for extracting advanced metadata from audio and video files, and Azure AI Content Safety to detect unwanted text or images.
Next steps
Account signup
Get started with a free account
Start with $200 Azure credit
Get started with pay-as-you-go pricing
There’s no upfront commitment—cancel anytime.
Get the Azure mobile app 
