[Release] How to access object-level caption and pose estimation from DINO-X?

[Everloom-129]

Hi IDEA team,

Thanks for the amazing work on DINO-X! I noticed that your paper and online demo mention support for multiple output heads including:

• Region-level/object-level captioning
• Pose estimation with keypoints for person/hands

Could you please clarify:

1. Are these modules accessible via DDS Cloud API?
2. If not yet public, is there a plan/timeline to release these features?
3. Are there any available examples or references to test these functionalities locally?

Really appreciate your work — this model has been very helpful for our robotics perception project!

Best regards,
Tony