- Newest
- Most votes
- Most comments
Amazon Bedrock Data Automation (BDA) does support custom output for images, but there are some important points to consider:
-
Custom output is available for both documents and images in BDA.
-
Within a single project, you can apply one image blueprint for custom output.
-
When creating custom blueprints for images, you define fields with descriptions, data types, and inference types. BDA supports various data types for images, including string, number, boolean, array, and inferred.
-
All image files processed by custom output must adhere to BDA's image processing restrictions.
If you're experiencing issues with custom output for images, here are a few things to check:
- Ensure your project is correctly configured with an image blueprint.
- Verify that your input image meets the BDA image processing restrictions.
- Check that you're using the correct API calls and parameters, including the proper Project ARN and Blueprint ARN.
- Review your S3 bucket permissions to ensure BDA has the necessary access to write the output.
If you've recently transitioned from the preview version to the generally available version, there might have been some changes in the service. It's recommended to review the latest documentation and update your code accordingly.
If you're still encountering issues after checking these points, it would be best to contact AWS support for further assistance, as they can provide more specific guidance based on your account and configuration.
Sources
How Bedrock Data Automation works - Amazon Bedrock
Creating blueprints for images - Amazon Bedrock
Custom output and blueprints - Amazon Bedrock
Hello,
Thank you for reaching out regarding the "type as image" modality in Amazon Bedrock. I understand you're experiencing difficulties with this feature, specifically in locating the custom output in S3. You've also mentioned that previously functional code, which worked during the preview state or with public access, is no longer operational. Please confirm if this accurately summarises your concerns.
I'd like to clarify that the custom results you're seeking are stored within the S3 bucket that was initially created when you set up your Bedrock Data Automation (BDA). To access these results, please navigate to the S3 bucket (inference_results/) and locate your results, organised by date and time.
In the event that you have deleted the initial S3 bucket created for your Bedrock Data Automation (BDA), please be aware of the following process when creating a new blueprint. Upon uploading an image during the blueprint creation process, you will encounter a bucket creation confirmation message. A "Create bucket confirmation" dialog will appear, which reads as follows: "Confirm creation of bucket with name: bedrock-bda-*******50798b0f4. This bucket will be used to store assets used with Bedrock Data Automation." To proceed, you must click "Confirm" to authorise the creation of a new S3 bucket. It's crucial to note that this newly created bucket will serve as the storage location for all subsequent results generated by your Bedrock Data Automation. Please ensure you keep track of this bucket name for future reference and result retrieval
Regarding the code-related issue you've mentioned, I'll need additional information to provide accurate assistance. I recommend opening a support case for this matter, as our dedicated support team will be better equipped to address your specific concerns.
For further guidance on using blueprints in the console, please refer to our documentation: https://docs.aws.amazon.com/bedrock/latest/userguide/bda-blueprints-console.html
Relevant content
- AWS OFFICIALUpdated a year ago
