We need animal pose estimation support in the ControlNet extension, similar to the existing human pose…
We need animal pose estimation support in the ControlNet extension, similar to the existing human pose…: a task in LegoFlow-SWE (Harbor dataset). Observed behaviour: - encode poses as json accepts only a list of human poses and canvas dimensions; it has no way to include animal keypoint data. -…
The task
**Observed behaviour:** - `encode_poses_as_json` accepts only a list of human poses and canvas dimensions; it has no way to include animal keypoint data. - The render endpoint returns an error or ignores any animal-specific JSON structure.
Part of Lego-X/LegoFlow-SWE.