> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/changelog/explore-by-month/june-2026/rf-detr-keypoint-preview.md).

# RF-DETR Keypoint Preview

RF-DETR Keypoint is now available in Roboflow as a preview. It is a real-time, end-to-end pose model that extends the RF-DETR family from boxes and masks to keypoints, released under Apache 2.0 for commercial use. You can label skeletons in Annotate, train on your own data in Train, and deploy with Inference and Workflows.

The model predicts a structured set of keypoints for every detected object in a single forward pass, with no NMS, heatmaps, or post-hoc grouping, and it learns calibrated per-keypoint uncertainty from your data. Skeletons are not limited to the 17-point human pose, so you can define any number of keypoints on any class for use cases like rep counting, ergonomic checks, sports analytics, robot guidance, and gauge reading.

[Read the launch post](https://blog.roboflow.com/launch-rf-detr-keypoint-in-roboflow/)