Document AI - specifying existing Google Photos albums and accounts as a Dataset for training?

[GeneAlchemist]

When training a custom processor, the only option for the Dataset seems to be a Google Cloud Storage endpoint. I already have my training images in Google Photos and would like to be able to point it to an existing album, account, or other structure.

Is there any way to accomplish this, besides having to manually copy over photos into a GCS location? That's very inefficient for my data set.

[kvandres]

Good day @GeneAlchemist,

Welcome to Google Cloud Community!

Unfortunately, you can't use google photos as a location for your dataset. You need to upload the documents first to the Google Cloud Storage Bucket and the processor will use that specific bucket to indefinitely import the documents in order to train or uptrain a processor version. you can check this link for more information: https://cloud.google.com/document-ai/docs/workbench/create-dataset

Hope this helps!