Optimal way to store data in Azure for use in Custom Vision

Henning Markussen 6 Reputation points
2022-09-22T07:21:32.027+00:00

We are gathering large amounts of data that we want to store on Azure. The purpose of the data is to be used to train ML models and we are also looking to do annotation with Azure using Custom Vision(CV). As of now we have uploaded it to Blob Storage and have developed python scripts to expose this data to CV.
This could suffice, but we are interested in finding a more optimal solution, and for this we have some questions.
When data is exposed to CV through ImageUrlCreateBatch, does it get stored in another place in Azure, or is it more like a mounting point? We would like to avoid double storing our data because of the environmental waste it incurs and the additional costs.
Would it be better to directly upload the data to CV? Can we then easily retrieve data to use in training, through either mounting points in AzureML or expose them to Datastores and Datasets for AzureML?
Any aide in developing a more efficient, cost effective and environmentally friendly solution is much appreciated :D

Not Monitored
Not Monitored
Tag not monitored by Microsoft.
35,529 questions
{count} votes