[Feature]: Addition of hand landmarks as skeletons #2170
Description
Search before asking
- I have searched the issues and discussions and found no similar feature request.
Feature Description
I'd like to propose adding support for MediaPipe Hand Landmarks as a new skeleton definition in supervision. Currently, the Skeleton class has support for COCO, GHUM, FACEMESH_TESSELATION_NO_IRIS and FACEMESH_TESSELATION.
The hand landmark definition has 21 keypoints per hand, so no structural changes would be needed, just a new enum member with 21 vertices, which doesn't overlap with any existing definition of SKELETONS_BY_EDGE_COUNT or SKELETONS_BY_VERTEX_COUNT. This alone would be very helpful just for annotation.
Additionally, the from_mediapipe() class method could be extended to handle this, requiring it to distinguish between the face_landmarks and hand_landmarks attributes.
Example Usage
No response
Are you willing to submit a PR?
- Yes I'd like to help by submitting a PR!