Skip to content

Pose Anything

Locate keypoints on any object category

Visit official site

What Pose Anything is

Pose Anything is described as a graph-based general pose estimation method that applies keypoint localisation to any object category.

What you can do with it

  • Locate keypoints on arbitrary objects
  • Use a single model for pose estimation
  • Work from a few annotated support images

Pros & cons

✓ What we like

  • Few annotated support images are needed
  • The material reports MP-100 benchmark results

! What to watch out for

  • Results depend on the annotated support images
  • The described method is a research approach on arXiv

FAQ

What benchmark is named?

The material names the MP-100 benchmark and 1-shot and 5-shot settings.

What makes it general?

The material says a single model handles keypoint localisation for any object category.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review