Easily customize your own state-of-the-art computer vision models for your unique use case.
Create a custom computer vision model in minutes
Customize and embed state-of-the-art computer vision image analysis for specific domains with Custom Vision, part of Azure Cognitive Services. Build frictionless customer experiences, optimize manufacturing processes, accelerate digital marketing campaigns, and more. No machine learning expertise is required.
Customization to your scenario
Set your model to perceive a particular object for your use case.
Intuitive model creation
Easily build your image identifier model using the simple interface.
Run Custom Vision in the cloud or on the edge in containers.
Rely on enterprise-grade security and privacy for your data and any trained models.
Achieve accuracy without the complexity
Start training your computer vision model by simply uploading and labeling a few images. The model tests itself on these and continually improves precision through a feedback loop as you add images. To speed development, use customizable, built-in models for retail, manufacturing, and food. See how Minsur, one of the world's largest tin mines, uses Custom Vision for sustainable mining.
Accelerate model creation
A user-friendly interface walks you through developing and deploying custom computer vision models. Then either ping the API to quickly tag images with your new computer vision model or export the model to a device to run real-time image recognition.
Deploy anywhere, from the cloud to the edge
Run your models wherever you need them and according to your unique scenario and requirements. Easily export your trained models to devices or to containers for low-latency scenarios.
Fuel App Innovation with Cloud AI Services
Learn 5 key ways your organization can get started with AI to realize value quickly.
Comprehensive security and compliance, built in
Microsoft invests more than USD 1 billion annually on cybersecurity research and development.
We employ more than 3,500 security experts who are dedicated to data security and privacy.
World-class custom computer vision at competitive prices
Pay only for what you use with no upfront costs. With Custom Vision, you pay as you go based on number of transactions, training hours, and image storage.
Get started with an Azure free account
After your credit, move to pay as you go to keep building with the same free services. Pay only if you use more than your free monthly amounts.
Documentation and resources
Frequently asked questions
Azure Cognitive Services, including Custom Vision, guarantees 99.9 percent availability. See SLA details.
Yes. Because Custom Vision is designed to be customized for your scenario, you need to provide the data to train your model.
The Custom Vision service is optimized to quickly recognize major differences between images, so you can start prototyping your model with a small amount of data. We recommend starting with 50 images per label. Depending on the complexity of the problem and degree of accuracy required, hundreds or thousands of samples may be required for your final model.
You can leverage Project Trove to gather images for your projects. Project Trove is an app that connects you directly with photo takers, allowing you to collect more relevant and accurate photos for your model. Using Trove, you can post your project descriptions, outline the types of photos you are looking for, and only approve the photos that you want. Trove provides licensing and privacy frameworks, so you can collect high quality data responsibly and safely. You can sign up for Trove here.
It’s actually both. Use the site to access a graphical interface for labeling data and training models. Or use the Custom Vision SDKs to do these things.
Cognitive Services offers several capabilities depending on your use case. The Computer Vision video- and image-recognition model is a good starting point if you don’t want to build your own model. Form Recognizer is optimized for document processing, Video Indexer for extracting advanced metadata from audio and video files, Face for facial recognition and detection, and Content Moderator for detecting unwanted text or images.