Image Classification: AWS -> MotherDuck with LaunchFlow

This is a simple example of how to use LaunchFlow to classify images as they are uploaded to an S3 bucket, and write the results to a table housed in MotherDuck.

To run this example, you will need to have an AWS account, and a MotherDuck account. You will also need to have the AWS CLI installed and configured with your AWS credentials.

Setup

In .env update S3_BUCKET_NAME to a bucket you would like to write the files in. This bucket does not need to be created yet, and update MOTHERDUCK_TOKEN to your MotherDuck service token.

Install Dependencies

pip install -r requirements.txt

Create your Resources

Be sure to run aws sso login before doing this!

Run:

buildflow apply

And type yes to confirm the resources you are about to create.

This will create a S3 bucket, a S3 Notification, and an SQS Queue.

Run the Example

Run:

buildflow run

Visit http://localhost:8000 for a simple UI for uploading images. Once uploaded you can view the results in MotherDuck: https://app.motherduck.com

Cleanup

If you would like to delete the resources created by this example, you can run:

buildflow destroy

and confirm the results. This will delete the bucket, notification, and queue.