This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(35.2, 69%, 13%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EAU%3C/text%3E%3C/svg%3E)
I'm having two different tenant in Azure.
I need to move all my custom vision model from one subscription to another subscription.
Hi @AzCredits User ,
to move Custom Vision Models you need to export and import the models:
The required project id you get this way: Get the project ID
(If the reply was helpful please don't forget to upvote and/or accept as answer, thank you)
Regards
Andreas Baumgarten
Only models trained using compact domain can be exported.
But all my models are General A1
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(166.4, 22%, 26%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ESC%3C/text%3E%3C/svg%3E)
Hello @AzCredits User,
Migrating Custom Vision models between two different Azure subscriptions across tenants is not supported through a native one-click feature in Azure. However, this can be accomplished by exporting and re-importing your projects and models manually or through automation.
It is recommended to back up your project metadata, including tags, iterations, and training images, using the Custom Vision Training API or the Azure SDK for Python. These backups help you replicate the original setup more accurately in the destination subscription.
Once you’ve set up new Custom Vision resources (Training and Prediction) in the target tenant and logged in to the Custom Vision portal, you can create a new project matching the original project’s configuration. If the model was exportable, you can deploy and use it directly for predictions either hosted on your own environment or using the exported package. If not, you'll need to retrain the model using the backed-up images and tags.
If you're working in an environment where both tenants or subscriptions are accessible via the same authenticated user or service principal, you may use the Custom Vision REST APIs (ExportProject and ImportProject) to perform the migration programmatically.
Use the ExportProject API to export your Custom Vision project from the source subscription. You'll need to find your training keys and endpoint URL from your Custom Vision resource for this step.
POST https://<your-resource-endpoint>/customvision/v3.0/projects/<project-id>/export
Once you have exported your project, you can then use the ImportProject API to import it into your target Custom Vision resource in the new subscription.
POST https://<your-target-resource-endpoint>/customvision/v3.0/projects/import
These APIs allow you to export a project from the source and import it into the destination, provided both resources support it and are configured correctly.
Please refer this for copying and backing up your Custom Vision projects here.
Before you begin the migration process, ensure that you have the necessary access to both Azure subscriptions involved both the source and the destination.
It's important to verify that you meet all prerequisite conditions, such as having the appropriate permissions on the Custom Vision resources.
Additionally, take time to familiarize yourself with the settings of your Custom Vision resources, as you will need specific details like the training keys and endpoint URLs in order to export, import, or retrain your models effectively.
I Hope this helps. Do let me know if you have any further queries.
If this answers your query, please do click Accept Answer and Yes for was this answer helpful.
Thank you!
I had trained the models using General A1 domain.
To export the model it needs to trained using Compact domain right?
Hi AzCredits User,
To export a model using the Azure AI Custom Vision Service, it must be trained using a compact domain. Since you mentioned that your models were trained using the General A1 domain, you will need to convert them to a compact domain before you can export them. After converting to a compact domain, you will need to retrain the model to ensure it is compatible for export.
Please refer this Export your model for use with mobile devices,
Select a domain for a Custom Vision project.
If this answers your query, please do click Accept Answer and Yes for was this answer helpful.
Thank you!
Hi AzCredits User,
Did you get any chance to check the above response, if this answers your query, please do click Accept Answer and Yes for was this answer helpful.
Thank you!