Keypoint Detection
Transformers
Safetensors
vitpose

vitpose-plus COCO-wholebody keypoints

#2
by TomVanWouwe - opened

My understandig was that setting the dataset_index to "5" would lead to getting the richer COCO-wholebody keypoints.
However, when setting dataset_index to "5", the output remains 17 keypoints. There also seems to be something wrong (see images).
Is this currently not supported or am I missing something?
dataset_index_0
dataset_index_5

Sign up or log in to comment