Forum Discussion

jortizflores's avatar
jortizflores
Icon for Microsoft rankMicrosoft
Nov 18, 2026

Azure ML Data Labeling retirement: rationale and recommended third-party annotation alternatives?

With Azure Machine Learning Data Labeling retiring on September 30, 2026, I'm interested in understanding both the rationale behind the retirement and what others are moving to as a replacement.

 

The official guidance recommends transitioning to third-party data labeling providers, but I'm particularly interested in computer vision workflows involving:

 

Image and video annotation

Bounding boxes and segmentation

Human-in-the-loop review

YOLO/COCO export

Azure Blob Storage integration

Private/self-hosted deployment

 

For those already evaluating alternatives, what have you found works well?

 

I'm currently looking at CVAT, Ultralytics Platform, Label Studio, and Roboflow, but I'd be interested in other recommendations and, especially, trade-offs you've encountered integrating them with Azure Machine Learning.

 

I'd also appreciate any insight into the product rationale behind retiring the first-party Azure ML labeling experience and whether there is a successor experience planned.

1 Reply

  • May be considering the potential replacement platforms:

     

    • CVAT: Best for self-hosted, video annotation, segmentation, and enterprise control.
    • Label Studio: Best for human review workflows and multi-modal labeling.
    • Roboflow: Best for rapid computer vision development and ease of use.
    • Ultralytics Platform: Best if your primary target is YOLO training.