With Azure retiring the Image Analysis 4.0 Segment API in early 2025, Florence 2.0’s “Region to Segmentation” within Azure Computer Vision is the recommended substitute.
If you're building a pipeline, you'll want to:
Call the new Computer Vision /analyze endpoint with the region-to-segmentation parameter.
Use the returned mask to composite a transparent-background image.
- Automate it via your existing codebase—Florence 2.0 supports standard API access just like the old Segment API
We’ve been experimenting with similar setups over at just build things ai image analyzer, working on end-to-end image-background removal using Florence. Happy to share sample Node.js or Python snippets if you're interested!